THE DETECTION OF PHARMACEUTICAL DRUG COMPOUNDS FROM INTACT
BIOLOGICAL TISSUE BY MATRIX- ASSISTED LASER DESORPTION
IONIZATION (MALDI) QUADRUPOLE ION TRAP MASS SPECTROMETRY
By
CHRISTOPHER D. REDDICK
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1997
To my parents
ACKNOWLEDGMENTS
Of all the people who have contributed to this work, no two people deserve more
credit than my parents George and Sue Reddick. Growing up, my parents never pushed
me to succeed or forced me down a path I did not want to travel. Instead they simply
loved me the best that they could and encouraged me in all of my endeavors big and small.
I am also thankfiil to my parents for being the best role models a boy could have. While
other kids were searching for role models in the movies or on TV, I only had to look
across the dinner table every night to find mine. Of all of the things my parents have done
for me over the years, their unconditional support has been their greatest gift. And for
that I will always be thankful.
At the University of Florida my sincerest thanks first go to my research advisor Dr.
Rick Yost for taking a young man with practically no mass spec experience into a research
group fiill of all-stars. Throughout my tenure in Rick's research group I could always
count on him for advice, feedback, and that extra shot of confidence when things weren't
going quite as planned. Most of all, I would like to thank Rick for allowing me the
freedom to work at my own pace and to really enjoy graduate school. I would also like to
thank the members of my committee, Dr. Dave Powell, Dr. Jim Winefordner, Dr. Jim
Deyrup, and Dr. Howard Johnson, for their time and effort in helping me complete this
dissertation.
HI
Probably the most difficult job I have ever undertaken has been designing and
constructing the instrument for this dissertation. No one person was more instrumental in
helping me accomplish this goal than Joe Shalosky, the Chemistry Department machine
shop supervisor. Joe is an expert machinist. But more importantly he is an excellent
teacher with endless patience. I would like to thank Joe for his time, effort, and most
importantly for the lively conversations we had during the long hours in the machine shop.
I would also like to thank Donna Balkcom for helping me navigate through the graduate
system at UF.
Without a doubt, every member of the Yost group has added in one way or
another to my experience at UF. I would like to first thank past Yost group members Uli
Bemier, Rafael Vargas, Jon Jones, and Tony Annachino for making the early years in
Rick's group fian. Thanks also go out to all of the members of the Burrito Brothers lunch
crew, past and present. No matter how bad the day was going, our daily trip to Burrito
Brothers always seemed to make things better. I especially would like to thank Scott
Quarmby for his help with the electronics for the instrument and in general for being a
fiiend. Working with Scott over the past four years has truly been educational. More
recently, I would like to thank Rick Troendle for being my co-pilot on the instrument
during my last few months of research. Rick's optimism and enthusiasm gave me a real
boost during the final push to graduate. The research for this dissertation was fiinded by
Bristol-Myers Squibb. I would like to thank Drs. Ira Rosenberg, Mike Lee, and Mark
Hail for their enthusiasm and support throughout the project.
Finally, I would like to thank my longtime fiiend and mentor at ALCOA, Robin
Khosah, for encouraging me to become a scientist and to get my Ph.D.
iv
TABLE OF CONTENTS
page
ACKNOWLEDGMENTS iii
ABSTRACT vii
1 INTRODUCTION 1
Development of Laser Desorption Mass Spectrometry 2
Matrix-Assisted Laser Desorption Ionization (MALDI) 6
MALDI Theory and Mechanism 12
MALDI Sample Preparation 14
MALDI Matrices 17
Overview of Dissertation 18
2 FUNDAMENTAL INVESTIGATIONS OF MALDI OF DRUG
COMPOUNDS IN TISSUE USING A TIME-OF-FLIGHT
MASS SPECTROMETER 22
Instrument Description 23
MALDI Optimization Experiments with Matrigel 25
Matrigel Sample Preparation 25
Optimization of the MALDI Matrix Concentration 27
Optimization of the MALDI Matrix Solvent Polarity 32
Optimization of the MALDI Matrix "Soak Time" 39
Spatial Resolution Experiments with Spiperone in Matrigel 40
3 DESIGN AND CONSTRUCTION OF A NOVEL LASER DESORPTION
QUADRUPOLE ION TRAP MASS SPECTROMETER 51
The Quadrupole Ion Trap Mass Spectrometer 51
Background History 51
Ion Trap Theory 56
Operation of the Ion Trap 63
Ion isolation 65
Tandem mass spectrometry (MS/MS) 70
Ion detection 70
Mass range extension 72
Coupling LDI to the Ion Trap 76
V
MALDI Inside the Ion Trap 76
MALDI Using an External Source Configuration 82
Instrument Design 92
Vacuum Manifold and Pumping System 92
Ion Source 95
DC Quadrupole Deflector Assembly 100
Laser Setup 102
Software Control 106
4 MALDI OF DRUG COMPOUNDS IN TISSUE USING A QUADRUPOLE
ION TRAP MASS SPECTROMETER 109
Instrument Calibration & Optimization 109
EI of Perfluorotributylamine Calibration Gas 109
Instrument Simulation using SIMION V6.0 115
High Mass Calibration using a Peptide Mixture 121
Analysis of Spiperone fi^om Rat Cerebral Tissue 122
MALDI MS and MS/MS of Standard Spiperone 125
Preparation of the Cerebral Tissue 128
MALDI Analysis of Cerebral Tissue 129
Analysis of Taxol from Mouse Ovarian Tumor Tissue 131
MALDI MS and MS/MS of Standard Taxol 132
Preparation of the Ovarian Tumor Tissue 137
MALDI Analysis of Ovarian Tumor Tissue 137
Analysis of Polymyxin Bi from Human Plasma 141
MALDI MS and MS/MS of Standard Polymyxin Bi 142
MALDI Analysis of Human Plasma 144
LD/CI as an Alternative to MALDI 147
Initial LD/CI Experiments with Trimethylphenylammonium bromide 151
LD/CI of Spiperone in Rat Cerebral Tissue 156
5 CONCLUSIONS AND FUTURE WORK 157
LIST OF REFERENCES 164
BIOGRAPHICAL SKETCH 172
vi
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
THE DETECTION OF PHARMACEUTICAL DRUG COMPOUNDS FROM INTACT
BIOLOGICAL TISSUE BY MATRIX- ASSISTED LASER DESORPTION
IONIZATION (MALDI) QUADRUPOLE ION TRAP MASS SPECTROMETRY
By
Christopher D. Reddick
August 1997
Chairman: Richard A. Yost
Major Department: Chemistry
The aim of the work presented in this dissertation was to investigate the use of
matrix-assisted laser desorption ionization (MALDI) for detecting pharmaceutical drug
compounds fi^om intact biological tissues. This research was also designed to evaluate the
potential of laser desorption ionization on a quadrupole ion trap for future laser
microprobe applications. Research efforts on this project were focused in three major
areas: fundamental studies, instrumentation, and applications.
In the first stage of the project, experiments were performed to evaluate and
optimize the MALDI process for drugs in a model tissue matrix (matrigel) using a
commercial MALDI-time-of-flight instrument. Two central nervous system drug
compounds, spiperone and ephedrine, were studied using the MALDI matrix 2,5-
dihydroxybenzoic acid. Results of these initial experiments showed that the concentration
of the matrix solution, the polarity of the matrix solvent, and the crystallization speed of
vii
the matrix were all important in increasing the production of analyte ion signal from the
model tissue.
In the second stage of the project, a novel laser desorption quadrupole ion trap
instrument was constructed for analyzing drug compounds from more complex tissues.
The instrument was constructed using a Finnigan series 4500 electron ionization/chemical
ionization source, a DC quadrupole deflector, and a Finnigan ITS40 quadrupole ion trap
mass analyzer. The advantage of this instrument over current laser microprobe instruments
is that it is capable of performing multiple stages of mass spectrometry (MS") for detecting
trace levels of analytes from complex tissues such as brain and liver. MS" can also be used
to determine the structure of drug compounds in tissue.
In the final stage of the project, the new instrument was used to analyze several
pharmaceutical drug compounds from tissue samples obtained directly from test species.
MALDI spectra were obtained for the antipsychotic drug compound spiperone from
incubated rat cerebral tissue. Experiments were also performed on samples of human
plasma spiked with the peptide antibiotic polymyxin Bi. Finally, MALDI was used to
detect the anticancer drug taxol in ovarian tumor tissue from mice that had been
administered the drug intravenously prior to removal of the tumor. The amount of the
drug compounds detected with the new instrument was determined to be in the low
picogram range.
viii
CHAPTERl
INTRODUCTION
In the past decade, research aimed at deciphering the human genome has rapidly
identified numerous disease targets. Using new sequencing methods and amplification
techniques such as the polymerase chain reaction (PCR), biologists and biochemist can
now identify disease causing mutations in specific regions of proteins and DNA.' The
natural outcome of this research has been an increase in the demand on the part of
physicians and healthcare providers for new therapeutic drug compounds and
medications.^ To meet this challenge, pharmaceutical companies have responded by
exploiting new analytical methodologies aimed at increasing the efficiency of their drug
discovery and development programs.^
One of the most important, yet challenging steps in the pharmaceutical drug
development process is elucidating the mechanism of drug action." Information about the
biological activity of drug candidates can be used to optimize a lead candidate and can
also give early clues as to a drugs metabolic pathway and possible toxicity. Traditionally,
drug action and metabolic profiling studies have been carried out on physiological fluids
(bile, urine, plasma) using standard techniques such as chromatography', nuclear magnetic
resonance spectrometry (NMR)*, fluorescence spectroscopy', and chromatography
coupled mass spectrometric techniques.* However, in order to fiilly elucidate the
mechanism of drug action at the cellular and subcellular levels it is necessary to know the
specific site of reactivity, or drug location in the body, as well as the chemical structure of
1
2
the drug compound at that specific site.^ To achieve these two goals analytically requires
techniques with high sensitivity, spatial and molecular resolution, and the ability to probe
into the subunits of biological matrices.'"
Development of Laser Desorption Mass Spectrometry
The combination of laser desorption ionization (LDI) and mass spectrometry is an
ideal technique for the analysis of biological materials. By using a focused laser beam,
specific regions of biological samples can be sampled with spatial resolution in the
submicrometer regime." In addition, the high energy deposition afforded by the focused
laser radiation (10^ - 10* W/cm^) allows for the vaporization and simultaneous ionization
of a wide range of thermally labile and nonvolatile biomolecules. LDI also has the
advantage over surface desorption techniques such as fast atom bombardment (FAB) and
secondary ion mass spectrometry (SIMS) of being able to probe through the cellular
matrix of biological samples.'^ And finally, LDI requires no separation and limited sample
preparation prior to analysis. Coupling LDI with mass spectrometry provides a sensitive
and selective technique capable of totally characterizing molecular species fi^om tissues by
determining their mass, relative abundance, and most importantly their structure (using
MS/MS techniques).'^
The development of laser desorption mass spectrometry began shortly after the
introduction of the laser in the early 1960s.''' The appeal of being able to rapidly heat a
solid sample with a coherent light beam to cause the ejection of electrons, neutrals and
especially ions led to an early marriage between the laser and mass spectrometer. The first
application of the laser in mass spectrometry was made in 1963 when Honig and
Woolston" adapted a pulsed ruby laser to a double-focusing mass spectrograph for the
elemental analysis of metal surfaces, semiconductors, and insulators.
Several years later Conzemius and coworkers'* revisited the technique. In their
work, a commercial laser was adapted to a double-focusing mass spectrometer of the
Mattauch-Herzog type, equipped with both electrical and photoplate detection. The
system had limited spatial resolution (-150 jim) and was used mainly for measuring
concentration profiles of trace elemental species in metal systems and thin films.
To date, LDI sources have been coupled with every type of mass analyzer design
including single and double focusing sector", quadrupole'*, time-of-flight'^, and ion trap
mass spectrometers.^" Probably the most widely sited laser desorption mass spectrometer
for biological applications is the Laser Microprobe Mass Analyzer (LAMMA).^' This
instrument, introduced commercially in 1977, is based on a time-of-flight mass
spectrometer and uses a Q-switched, fi-equency-multiplied Nd:YAG sampling laser, a He-
Ne spotting laser, and a modified laser focusing microscope system capable of achieving
spatial resolution as low as 0.5 iim (Figure 1-1). The LAMMA was originally developed
for high sensitivity molecular analysis of thin histological sections, but has been primarily
• ■ 22 25
used for in situ determination of physiological cations m organ tissues.
Vandeputte and Savor^* used the LAMMA to study the localization of aluminum
in livers of aluminum maltol treated rabbits. This model was developed to study long-term
aluminum toxicity. In their work, aluminum maltol was administered to adult male rabbits
intravenously three times a week for 8-30 weeks. Liver sections were fixed in 10%
buffered formalin, embedded in paraffin, and mounted on copper grids. While the
LAMMA was not used to image aluminum in the liver sections, significant accumulations
4
5
of aluminum were detected in electron-dense deposits in cells found in pathological
lesions of the liver tissue. Improving on this work, Verbueken and coworkers^* used the
LAMMA to analyze kidney sections from rats treated with the immunosupressive drug
cyclosporin. Instead of looking for metabolites of cyclosporin in the kidney sections, the
LAMMA was used to pin-point and determine the chemical composition of numerous
intrarenal crystalline deposits. Results of this work revealed the microcrystalline
structures to consist of calcium oxalate, sodium urate, and calcium phosphate. Similar
deposits in human kidney sections were also characterized by the LAMMA.
Applications of laser desorption mass spectrometry for the analysis of molecular
species from biological matrices have not been widely reported in the scientific literature.
This is due primarily to the fact that many biomolecules of interest undergo thermal
degradation at the high laser irradiances needed to vaporize and ionize regions of tissue.^'
The laser desorption of molecular species from tissues occurs both resonantly and
nonresonantly.^" In resonant desorption, the direct resonant excitation of the analyte
molecules channels energy into vibrational modes which can lead to photodissociation. In
the case of nonresonant desorption, the high irradiances required for desorption occur
very close to the point of plasma generation, which can also lead to molecular
decomposition. In either case, the intense energy deposition from the incident laser
irradiation destroys the molecule.^* The problem of performing laser desorption out of
tissues is further complicated because the traditional laser microprobe instruments do not
have the selectivity and sensitivity to detect analyte fragments in the presence of the
intense background noise from the tissue.^^
6
Matrix- Assisted Laser Desorption Ionization (MALDI)
In 1985, Hiilenkamp and Karas^^ reported laser desorption of intact (M+H)"^ ions
for the dipeptide, Trp-Trp (M.W. 390) using a pulsed nitrogen laser. In this work, a low
concentration of the analyte was mixed with a liquid matrix consisting of a low molecular
weight, UV-absorbing compound. It was observed that a strong resonance absorption of
the matrix compound at the wavelength of the incident laser radiation promoted a soft
desorption, at low laser irradiance, of the dipeptide without fi-agmentation. Expanding on
this pioneering work, Hiilenkamp and Karas went on to demonstrate the production of
intact molecular ions for several proteins with masses up to 67,000 Da (Figure 1-2).^''
Matrix-assisted laser desorption ionization or MALDI, as the technique is now termed,
revolutionized the field of laser desorption mass spectrometry by providing a means of
producing intact molecular ions for thermally labile biomolecules.
Since its introduction, MALDI has been used primarily to study neat samples of
biomolecules including peptides, proteins, glycoproteins, glycosides, nucleosides, nucleic
acids, and oligosaccharides with masses between 10,000 to 300,000 Da."'^* In the work
presented here, MALDI was used to produce intact molecular ions for small
pharmaceutical drug compounds (300 - 1200 Da) directly fi-om intact biological tissues.
While the literature is lacking in this specific area, there have been reports of similar uses
of MALDI for detecting molecular species embedded in other types of solid materials.
Hercules and coworkers^^ used MALDI to detect various compounds directly from
polyamide thin layer chromatography (TLC) plates. The compounds analyzed included
polyaromatic hydrocarbons (PAHs), alkaloids, and amino acids. The MALDI spectra
7
8
were obtained using a Q-switched, pulsed Nd:YAG laser (265 nm) with a power output
of 18 ^J and a spot size of ~5 |im. Samples were prepared by spotting 0.2 ^iL of the
sample mixtures onto TLC plates so that approximately 50 pg of each compound was
deposited. Separations were performed using methanol/water and methylene
chloride/methanol as the mobile phases. After the separation, the plates were allowed to
dry at room temperature and the spots were visualized with UV fluorescence. MALDI
was performed by depositing a drop of concentrated matrix solution directly on top of
each of the separated spots. Two matrices were evaluated; DHB and sinnapinic acid.
Of particular note in this study was the detection of the antibacterial drug
compound erythromycin (M.W. 735).^^ Abundant (M+H)* and (M+Na)"^ ions were
obtained for MALDI of erythromycin from the polyamide TLC plates. Background ions
from the TLC plate were also observed below m/z 150, but were found not to interfere
with the molecular ion species observed. Higher laser irradiances were needed to obtain
MALDI spectra from the TLC plates than from a standard metal substrate. This was
believed to be due to a combination of scattering and absorption by the TLC plates. Also,
reducing the plate thickness from 250 ^im to 100 \im was found to increase the production
of molecular ion signal for all of the compounds studied, including erythromycin.
More recently, MALDI has been applied to the analysis of proteins separated by
gel electrophoresis and electroblotted onto membranes. Gels currently in use are too
fragile for most manipulations, and it is increasingly common for separated proteins to be
electroblotted onto more robust polymer membranes by the application of an orthogonal
electric field. ''^ Several groups have reported MALDI of proteins from poly(vinylidene
difluoride) (PVDF) membranes up to 67,000 Da. Vestling and Fenselau'" have
demonstrated the usefulness of MALDI for providing molecular weights for several
proteins, including horse heart cytochrome c, lysozyme, and bovine trypsin using a-cyano-
4-hydroxycinnamic acid (M.W. 171) as the UV-absorbing matrix. In this work, the
proteins were transferred to PVDF membranes, washed with water to remove
contaminants, and allowed to dry. 1.0 ^iL of the matrix solution (100 mM in 50:50
methanol/toluene) was then applied to each spot on the membrane. Spectra were obtained
by scanning across the membranes with a focused beam from a pulsed 337 nm nitrogen
laser (Figure 1-3).
Hillenkamp and coworkers'** used infrared MALDI at 2.94 ^m to desorb proteins
directly from PVDF, polypropylene (PP), nitrocellulose, and polyamide blot membranes.
A variety of methods were employed to add the matrix to the blots. Soaking the dried
membranes in organic solutions containing the matrix compound (succinic acid) was found
to provide the most intense molecular ion signals for all of the proteins tested. Membrane
thickness and surface area were also found to influence the production of protein signals.
The results obtained for IR-MALDI were compared v^th those obtained by UV-MALDI
at 355 nm using DHB as the matrix (Figure 1-4). In all cases, IR-MALDI showed
superior results. The increase in molecular ion signal with JR desorption versus UV
desorption was believed to be related to the different penetration depths of the laser
radiation at the two wavelengths (3-5 [im vs. 200-300 nm, respectively).
CYTOCHROME C
TRYPSIN
INHIBITOR
CARBONIC
ANHYDRASE
BOVINE
ALBUMIN
A MIXTURE OF
THE ABOVE
I
f
9
100
90-
80
70
GO
SO
40
30
20
10
123^
CytochromeC
29050
Anhydrase
66332
Bovine
A&Hjmln
10000
20000
30000
40000
50000
60000 70000
Mass (m/z)
Figure 1-3.
MALDI spectrum of Cytochrome C, Trypsin Inhibitor, Carbonic
Anhydrase, and Bovine Albumin electroblotted onto a PVDF
membrane using a-cyano-4-hydroxycinnamic acid."*'
11
« 100 !
I 75
^ 50
I 25
10000 40000
m / z
10000
m / z
• • • •
Figure 1-4.
Comparison of UV (top) and IR (bottom) MALDI spectra of Soybean
Trypsin Inhibitor (M.W. 19,979) electroblotted onto a PVDF membrane.
Both spectra were obtained using saturated DHB in ethanol.^
12
MALDI Theory and Mechanism
Although significant work has been done in the area of MALDI, the actual
mechanism of how MALDI works is still not fully understood/*' In general, MALDI
involves rapidly depositing energy into a solid lattice of analyte embedded in matrix
crystals using short (3-300 ns), intense (10^ W/cm^) laser pulses. Part of the deposited
energy is reemitted through fluorescence. Another portion is channeled into vibrational
modes of the matrix molecules. Some of the matrix molecules decompose from these
vibrational states; others transfer their energy to the crystal lattice causing rapid heating to
the phase-transition temperature. An expanding gas-phase plume is then formed
containing highly excited matrix molecules with entrained analyte molecules (Figure 1-5).
It has been suggested that the lack of degradation of the analyte molecules during the
initial laser desorption event is due to a frequency mismatch between the lattice vibrations
in the matrix crystals and the intermolecular vibrations in the encapsulated analyte
molecules."*
The production of analyte ions in MALDI can be divided into at least three
different processes'*', which depending on the wavelength and irradiance, contribute to a
varying extent to the overall ionization of the analyte. In the first process, preformed
analyte ions in the solid matrix are volatilized into the gas phase during the initial laser
desorption event. Analyte molecules in their ionized form in the condensed-phase have
been shown to give rise to gas-phase ions extremely easily. The presence of abundant
(M+Na)* and (M+K)"^ ions in MALDI are thought to result from preformed adducts
between the analyte molecules and salts from the matrix solvent.'" The second process
® ®®^0® Q ®®00®®®®®O©®®®
• O 7^^^^^ ® ® O 0^,^i^® ® O O G O O O
® ® ®® /ooooeoo®
O®® ®® ®® ®®G®<^v^® 9®® ®® ®
o®o®®oooeo®yy® ® ®®^,SL^
9®eo®o&9®®®«^^ ®®®r09
®og®®q/\j ®
GOG OGdT/^® O ® ^^^^^"^-^ ®
0®G®®^J® G®®®®®0®G ®0®
9®®G®0®®®G®®®®®®®®®®
® ®0 GOO® ®0 0 000® ®®0 Q® ®
i
hv
®
®^ ® ®©© ©
® 0 © ^ ^ G
® ® ® © © ® ®
® c^^*
® © ®^® ®
® ® ® ® © o • ©•
©® G® C(«»® GGO
• G O® ® © S J ® OGGO
®GO ®® GGGGG
©GO®®© ® ®®®GO
0®G®®®®®©®© ®®G®G ^ S^©
®®G©®®P^©® ©©®® ®®®p09
®OG®®®/\j ® © ©®®©©©$lI (W9
© O G ® /© ® © © G © © O ©dx..^ G
©©®®®^_J® GOOG®®®®®®®®
©®®G®®®©®®®©®®©®®®®®
®®®G®®®®Q®®©®®®®©®®©
- analyte ® - matrix • - cation
Figure 1-5. Diagram showing the formation of the supersonic gas-phase plume of
matrix and analyte molecules during MALDI.''
14
involves direct photoionization and photofragmentation of individual molecules (photon-
molecule interactions, comparable to gas-phase photoionization). This process plays a
increasingly important role at higher laser irradiance. The final ionization process involves
protonation reactions by collisions of highly excited matrix ions and ground-state analyte
molecules. As one of the elemental steps of energy transfer fi-om matrix to analyte
molecules, the formation of intermediate radical ions has been proposed.
M"^ M'"^ + e" (1-1)
These radicals are photoionized in the condensed phase and react with ground-state
analyte molecules in the dense gas-phase plume formed just above the solid surface,
resulting in subsequent ionization by proton transfer reactions.
M "" + A ^ (A+H)^ + (M-H)' (1-2)
Rapid adiabatic cooling in the expanding jet prevents the intact analyte ions from
fragmenting.
MALDI Sample Preparation
One of the advantages of MALDI is that the preparation of samples is simple, fast,
and requires relatively little material. Traditionally, a 5-10 mg/mL solution of the matrix
material is prepared in water or water/organic solvent. For analysis of polar biomolecules,
methanol, acetone, and acetonitrile have been used. It is important to note that the
selection of the solvent system can have a dramatic effect on the crystal formation and
therefore on the quality of the resulting MALDI spectra. A suitable amount of the analyte
15
solution (10"^ - 10"* M) is then mixed with 5-10 //L of the matrix solution to yield a molar
ratio of 10^ - 10* (matrix-to-analyte). For large peptides and proteins, 0.1%
trifluoroacetic acid is added to aid the solubilization of the analyte." For analysis, a drop
(0.5-1.0 |iL) of the matrix/analyte mixture is then applied to the sample probe (typically
stainless steel or copper) and allowed to dry at room temperature. Passing a stream of
warm air or nitrogen over the droplet to speed crystallization of the sample is sometimes
performed."
Several improvements in sample preparation have been reported for MALDI with
the aim of increasing the sensitivity, shot-to-shot-reproducibility, and resolution of the
technique. Vorm and Mann" described a sample preparation procedure which decouples
matrix and sample handling. In this technique, the matrix solution is applied to the sample
probe in a solvent that evaporates very rapidly, leaving a thin layer of very small matrix
crystals. A small volume of analyte solution (0.3-1.0 |iL) is then added on top of the
matrix surface and allowed to dry. An alternative to this approach is to electrospray the
matrix/analyte solution as a fine mist by applying a small potential between the sample
syringe and the sample probe. Capriolii et al.'"* demonstrated MALDI of the
neurotransmitter Substance P fi^om dialysis probes using this technique. With both
techniques, increases in sensitivity (subfemtomole) and TOF mass resolution were
reported, believed to result fi^om the very flat, homogeneous crystal layers that formed.
Hercules et al." showed that the choice of matrix compound and the speed of drying of
the matrix/analyte mixture has the most pronounced affect on the crystal structure and
therefore on the quality of the resulting spectra (Figure 1-6). In their work, an increase of
Figure 1-6. Scanning electron images showing the crystal formations for various
matrices and drying speeds: Ferulic acid (top left), Ferulic acid with
accelerated drying (middle left), Ferulic acid/Frucose with accelerated
drying (bottom left), DHB (top right), DHB with accelerated drying
(middle right), and DHB/Fucose/5-methylsalacylic acid (bottom right)
17
ca. 50% in signal intensity and a 30-40% increase in resolution were obtained using DHB
with accelerated drying.
For MALDI of drug compounds from intact tissue in this dissertation research,
both the standard droplet method and the electrospray deposition method were used.
Descriptions of the sample preparation procedures for these experiments are presented in
chapter 2. A possible alternative for performing MALDI of biological tissues is to freeze
the sections and use the ice crystals as the absorbing matrix. Williams and coworkers'^
have reported MALDI from frozen matrices using an IR laser (266 nm) for the analysis of
single- and double-stranded DNA up to 29,000 Da. Samples were cooled to ~ 253K using
a liquid nitrogen cold finger attached to a copper sample stage.
In general, the two most important factors in preparing samples for MALDI are
analyte/matrix solubilization and crystal formation. Regardless of the technique chosen,
the analyte must be made to dissolve in the matrix solvent and the resulting crystal lattice
that forms upon drying must be flat and homogeneous in order for quality MALDI spectra
to be obtained.
MALDI Matrices
The basic prerequisites for a compound to work as a matrix in MALDI are as
follows.*' First, the compound must exhibit a strong spectral absorbance at the
wavelength of the incident laser radiation. For most of the MALDI experiments reported
in the literature, UV lasers including N2 (337 nm)", eximer (193, 248, 308, and 351 nm)",
frequency-doubled eximer-pumped dye (220-300 nm)", and Q-switched, frequency-
tripled and quadrupled Nd:YAG (355 and 266 nm, respectively)'* have been used. IR-
18
MALDI has also been reported using TEA-Co2 (10.6 fim) and Er:YAG (3 fxm) lasers.''
The second requirement is that the matrix compound must be chemically compatible with
the analyte of interest. Generally, this means that the matrix compound should be soluble
in the same solvents as the analyte. The matrix compound should not however undergo
any sort of chemical reaction with the analyte in solution. Finally, matrices should be
acidic so as to promote ionization of the analyte molecules via proton-transfer reactions.
Russell and coworkers*" investigated the affect of matrix acidity on the production of
analyte (M+H)* ion yield using a series of /^-substituted anilines as the matrices. Results
of their work showed a linear increase in ion yield with increasing excited state acidity of
the matrix compound.
Matrices in MALDI can generally be classified based on their absorbance
wavelength and on their application to either biomolecules or more recently synthetic
polymers. A list of current matrices and their use in various applications is given in Table
1-1. For all of the work presented in this dissertation, 2,5-dihydroxybenzoic acid (DHB)
was used as the MALDI matrix compound.
Overview of Dissertation
Research efforts on this project were focused in three major areas: fundamental
studies, instrument development, and applications. In the first stage of the project
experiments were performed to optimize the MALDI process using matrigel, a model
tissue matrix. The goals of these initial experiments were to elucidate the mechanism of
MALDI for drug compounds in tissue and to determine the effect of MALDI on the
spatial distribution of drug compounds in tissue. MALDI was demonstrated for the
19
Table 1-1 . Typical MALDI matrices and their use in various applications.^'
Matrix
Application(s)
^ra«^-2,5-Dimethoxycinnamic Acid
Higher mass biopolymers, glycoproteins, peptides, polymers
a-Cyano-4-hydroxycinnamic Acid
Organic compounds from 200-1000 Da, glycoproteins
2,5-Dihydroxybenzoic Acid
Polymers, polypeptides, oligosaccharides, glycopeptides
2-{4-Hydroxyphenylazo)-benzoic
Low molecular weight compounds, sulfonic acids, dyes
ZjHjO" 1 luiyuiUAytiwcLupiiciiuiic
OlionfiiiplpntiHpQ nolvmf*rc Hinnnl vmprQ
RinnolvmprQ nliQAniir'lftntiHf**! maQ^ ranof* 400-^^0 000 Oa
, \} LJ li 1 y \ii u Ay ctv w lupi 1^1 ivji It-
RinnnlvtrifTQ nlianniif'Ipi^tiHf*^ ma^^ ranffp 400-^0 000 Da
S-r^filnrnQjilifvlir AriH
\Vater irmoluble nolvmers
S-\4^f*tlinYVQjilif*vlir* A fin
TI^f*iH in ri^mhinatinn with Ol-TR for hpttf*r rp^nliitinn
wS^U 111 wV^lllL/llld-LlV/lt WlLll L XV/1 L/Vll'^l I V<9V/1UL1V7I1
1 R 0-T'rilivHrr»vvjint}irnr'i*np
l^^ll L/VICU WV^lllLIV^UllUO <U1U LyUiyillwlo
PnlvtlivmiHinpQ anH nnlviiriHinpQ
IT Lily lliyilLlUlilwO (UlU LAJiy Ul lUUlVo
Indoleacetic Acid
Oligonucleotides
N-(3-Indoleacetyl)-L-leucine
Oligonucleotides, polycytidines, polyuridines
Anthranilic Acid
Oligosaccharides, glycopeptides, glycoproteins
3-Hydroxypicolinic Acid
Oligonucleotides, polycytidines, polyuridines
Nicotinic Acid
Proteins and peptides
Vanillic Acid
Proteins and peptides
Pyrazine-2-carboxylic Acid
Proteins and peptides
3-Aminopyrazine-2-carboxylic
Acid
Proteins and peptides
Ferrulic Acid
Proteins, peptides, amino acids
CafFeic Acid
Proteins
3-Nitrobenzyl alcohol
Used for FAB and flow MALDI
Nitrophenyl octyl ether
Polymers up to 10 IcDa (salt)
20
central nervous system drug compounds spiperone and ephedrine from matrigel using a
Finnigan MAT Lasermat MALDI-time-of-flight instrument. The matrix concentration,
matrix solvent polarity, and the "soak time" of the matrix solution were found to be
important in the production of analyte molecular ion signal.
Spatial resolution experiments were also performed with spiperone in matrigel.
Results of these experiments revealed that addition of the MALDI matrix solution as a
droplet caused significant migration of the analyte. An electrospray apparatus was
constructed and used to spray the matrix solution on top of the tissue as a fine mist. With
this method, the migration of spiperone in matrigel was prevented. The results of this
initial work are presented in Chapter 2.
Chapter 3 describes the design and construction of a novel laser desorption
quadrupole ion trap mass spectrometer designed specifically to detect and potentially map
trace levels of drug compounds in complex tissues such as brain, liver, and tumor tissue.
The instrument was constructed using a modified Finnigan MAT model 4500 electron
ionization (EI)/chemical ionization (CI) ion source and a Finnigan ITS40 quadrupole ion
trap mass analyzer. A DC quadrupole deflector was incorporated to allow the ion trap to
be positioned 90° ofF-axis with the ion source. This configuration was used to allow light
from a nitrogen laser to be directed perpendicularly onto the surface of the sample held
within the ion source. The ofF-axis design will also allow a microscope objective to be
incorporated for future imaging experiments. Included in chapter 3 is an introduction to
the quadrupole ion trap mass spectrometer, complete with a historical review and
discussion of its theory and operation.
21
Chapter 4 presents the application of the new instrument for the analysis of drug
compounds in tissue obtained from test animals. The chapter begins with the calibration
experiments performed using perfluorotributylamine (FC43) and a peptide mixture.
Simulation experiments were also performed with SEMION V6.0 to optimize the various
instrumental parameters for MALDI. In the first application experiment, spiperone was
detected from a thin section of rat cerebral tissue using DHB as the MALDI matrix. The
sample was prepared by incubating the tissue section in a 10"^ M solution of spiperone.
Resuhs of these experiments showed that the MS/MS capabilities of the ion trap were
necessary to confirm the presence of spiperone in the more complex brain tissue. In the
second experiment, MALDI was performed on thin sections of rat ovarian tumor tissue
containing the anticancer drug taxol. In contrast to the previous experiments in which the
drug compound was mixed with the tissue after it had been removed fi^om the animal,
taxol was injected directly into the rat approximately 1 hr before being sacrificed. Based
on the initial loading of taxol in the tumor, the spot size of the laser, and the thickness of
the tissue section, the amount of taxol detected was determined to be approximately 280
pg. MALDI MS/MS spectra were also obtained for the peptide antibiotic polymyxin Bi
mixed with human plasma.
Chapter 5 concludes the dissertation with a discussion of the results obtained fi'om
the experiments presented in this work. Also included in this final chapter are suggestions
for instrumental improvements and a discussion of fiiture applications and experiments.
CHAPTER 2
FUNDAMENTAL INVESTIGATIONS OF MALDI OF DRUG COMPOUNDS IN
TISSUE USING A TIME-OF-FLIGHT MASS SPECTROMETER
In the first stage of this project, experiments were performed to investigate the use
of MALDI to detect drug compounds fi"om a model tissue matrix. The goals of these
initial experiments were threefold: to elucidate the mechanism of MALDI for drugs in
tissue, to optimize the MALDI process to increase the production of analyte molecular ion
signal, and to evaluate the usefulness of MALDI for potentially imaging drug compounds
fi-om more complex biological matrices.
To avoid using large quantities of tissue fi-om test animals for these initial
optimization experiments, a commercially available model tissue matrix (matrigel), was
used instead. Matrigel (Collaborative Biomedical Products) is a collection of extracellular
membranes underlying cells in vivo extracted fi-om Engelbreth-Holm- Swarm (EHS) mouse
sarcoma.*^ Matrigel' s minor components are laminin, collagen type IV, heparin sulfate
proteoglycans, entactin, nodogen, tissue plasminigen activator, and other naturally
occurring growth factors.^^ Matrigel proved to be an ideal model tissue matrix for these
experiments because of its unique physical properties. In its fi-ozen storage state matrigel
was a solid. As it warmed to room temperature it became a liquid, enabling drug
compounds to be easily mixed in with it. At room temperature the matrigel/drug mixture
solidified, locking the embedded drug compounds into position.
22
23
Instrument Description
The instrument used for this work was a Finnigan MAT Lasermat MALDI-time-
of-flight (TOF) mass spectrometer (Hemel Hemstead, UK) (Figure 2-1). The Lasermat
was used to perform the initial optimization experiments while the laser desorption ion
trap instrument was being constructed. Although the Lasermat was originally designed to
perform routine analysis of peptides and proteins,*'* it required no modifications for
MALDI of drugs in tissue.
With the Lasermat, light fi-om a pulsed nitrogen laser (337 nm) was used to desorb
ions fi"om samples deposited onto a 35 mm stainless steel sample plate having a target area
2.0 mm in diameter. The nitrogen laser produced 3 ns pulses of approximately 100 in
energy. Although the laser can fire at a repetition rate of 20 Hz, the system was operated
at 1 Hz. The laser beam was focused down to a spot size of approximately 0.1 mm by 0.3
mm using a single fused-silica lens (50 mm focal length). The laser power density was
adjusted fi-om 10* - lO' W/cm^ using a rotating polarizer under computer control. Control
of the laser beam aim was achieved using a rotating fused-silica wedge.
The ion source of the Lasermat consists of five stainless steel lenses which were
used to direct and accelerate the desorbed ions down the length of the 0.5 m flight tube.
Very high potentials (± 20 kV) and small distances between the lenses were maintained to
ensure that ions were ejected into the flight tube at the same time. Detection of the
desorbed ions was achieved with a discrete dynode electron muhiplier situated at the end
of the flight tube. Mass determination was made by measuring the flight time (t) of the
25
desorbed ions of given mass-to-charge (m/z) down the flight tube of known length (1),
after being accelerated in the extraction field to a common energy (E).*'
m t ^
- = 2E- (2-1)
z 1
MALDI Optimization Experiments with Matrigel
Matrigel Sample Preparation
The test compounds chosen for this work were the central nervous system (CNS)
drug compounds spiperone (M.W. 395) and ephedrine (M.W. 165) (Figure 2-2).
Spiperone and ephedrine proved to be ideal model compounds for this initial work
because they were both soluble in matrigel and because they have been studied extensively
in previous work in our laboratory.^^** Standards of spiperone in matrigel (100 ppm)
were prepared by mixing 0.1 mg of solid spiperone (Sigma Chemical Co.) with 1.0 mL of
matrigel after it melted from its frozen storage state. The mixture was vortexed until all of
the spiperone had visibly dissolved in the matrigel. After mixing, the spiperone-matrigel
mixture was allowed to gel at room temperature. Standards of ephedrine in matrigel (100
ppm) were prepared in the same manner by mixing 0.1 mL of a standard solution of
ephedrine (1 mg/mL) (Alltech Associates, Inc.) with 1.0 mL of matrigel.
26
H CH3
HO — C CH— N— H
b) M.W. 165
Figure 2-2. Structures and molecular weights of the central nervous system (CNS) drug
compounds studied: a) spiperone and b) ephedrine.
Optimization of the MALDI Matrix Concentration
27
Before analyzing the matrigel samples by MALDI, LDI reference spectra were
acquired for spiperone and ephedrine in matrigel without the addition of a matrix solution
(Figure 2-3 and Figure 2-4). The LDI spectra for both compounds were dominated
primarily by ions from the matrigel tissue. The presence of these background ions
suggested that some of the components of the matrigel tissue absorbed strongly in the
ultraviolet region. Most of the background ions were identified as low molecular weight
amino acids from the collagen in the matrigel. Abundant sodium and potassium ions were
also detected due to the salt content of the matrigel tissue. For the matrigel sample
containing spiperone, (M+Na)"^ and (M+K)* adduct ions were also detected. The lack of
abundant (M+H)"^ ions for either spiperone or ephedrine with laser desorption alone was
due to the inability of the laser to desorb and ionize those analyte molecules located below
the tissue surface. Higher laser powers were employed to try to probe into the matrigel
tissue. However, no (M+H)"^ ion signal was detected even at the maximum laser power
setting of the instrument. After analysis, observation of the matrigel surface under
magnification revealed no evidence of laser ablation or crater formation. No attempt was
made to focus the laser beam more tightly to provide higher irradiance levels because of
the complicated design of the Lasermat optics system.
After trying LDI unsuccessfially, MALDI was performed using 2,5-
dihyroxybenzoic acid (DHB) as the UV-absorbing matrix (Figure 2-5). In the first
experiment, 1.0 joL of the matrigel standard containing spiperone was deposited onto the
center of a Lasermat sample plate using an Eppendorf micro-pipette and allowed to air dry
28
100% « 1569
,00 Na* K +
90-
SO-
TO-
J 60-
I
S 40-
(A
C
0)
a:
30-
20-
10-
(M+H)+
m/^396
(M+Na)+
'mAE418
(M+K)+
'm/z434
I
50
100 150 200 250 300 350 400 450 500 550 600
100% « 3510
100
80
8&
<4->
w
70-
c
o
Int
60
ive
50-
40-
0>
30-
20-
10
(M+H)+
m/z396
DHB Matrix Ions
nn/z137
m/z155
100 160 200 250 300 350 400 450 600 550 600
MsM(in/z)
50
Figure 2-3. Comparison of the LDI spectmm (top) and the MALDI spectrum (bottom)
using DHB (0.25 M) for 100 ppm of spiperone in matrigel.
29
100%a1618
Na+
100
80-
'(0
70-
0)
60-
c
>
50-
(0
40-
"S
30-
20-
10-
K +
n
350 400 450 500 550 600
Mass (m/z)
— I —
SO
100
150
200 250 300
100%b27BO
100'
in/z137
m/z 155
m/i177
Na+
(M+H)+
m/z 166
(M+K)+
m/z 206
— I —
50
100
150
200 260 300 360
400 450 500 550 600
Mass (m/z)
Figure 2-4. Comparison of the LDI spectrum (top) and the MALDI spectrum (bottom)
using DHB (0.25 M) for 100 ppm of ephedrine in matrigel.
30
Figure 2-5. Structure of the MALDI matrix compound 2,5-dihydroxyben2oic acid
(DHB), M.W. 154.
31
at room temperature. The amount of drug compound deposited was determined to be 100
ng. With larger amounts of matrigel (>3 |iL), the sample did not dry completely. After
drying, the sample was washed with successive drops of water to remove any undisolved
analyte molecules from the tissue surface. 1.0 of a 0.01 M DHB matrix solution
prepared in 30% acetonitrile/70% water was then added directly on top of the sample.
Upon drying, the DHB matrix solution crystallized on top of the matrigel, forming needles
of several hundred micrometers in length pointing inward from the rim of the sample area.
While the first sample continued to dry, two more matrigel samples containing spiperone
were prepared for MALDI using the same volume of DHB matrix solution (1.0 |iL), but
at increasing concentrations of 0.15 M and 0.25 M. The same sample preparation
procedure was repeated for the matrigel samples containing ephedrine.
As can be seen in Figures 2-3 and 2-4, switching from LDI to MALDI resulted in a
dramatic increase in the production of the (M+H)"^ peak for both spiperone and ephedrine
in matrigel. For spiperone, the area of the (M+H)* peak increased from 400 using the
0.01 M DHB matrix solution to 3439 using the most concentrated matrix solution. For
ephedrine, the (M+H)^ peak area increased from 349 using the 0.01 M solution to 2363
with the 0.25 M DHB matrix solution. The MALDI spectra for the matrigel samples also
included several intense peaks corresponding to the DHB matrix. The two most abundant
peaks corresponded to the (M+H)^ ion at m/z 155 and for the (M+H - H20)^ ion at m/z
137. Less intense peaks at m/z 177 and m/z 274 were also seen for the (M - H2O + K) *
ion and the (2M - 2H2O + U)* ion respectively.
The increase in the (M+H)* peak area observed for both compounds using MALDI
is believed to have resulted from a two-step process. In the first step, the embedded
32
analyte molecules were extracted out of the matrigel and into the solvent of the DHB
matrix solution (Figure 2-6). In the second step, the extracted analyte molecules became
encapsulated in the matrix crystal layer that formed on the surface of the matrigel as the
solvent evaporated. Increasing the concentration of the matrix solution further increased
the production of (M+H)"^ ions by providing an excess of surrounding matrix molecules to
promote the desorption and protonation of the extracted analyte molecules.
In addition to increasing the production of the (M+H)* ion signal, addition of the
DHB matrix was also found to reduce the sodium and potassium adduct ions and suppress
the background ions from the matrigel which complicated the spectra. Under
magnification it was observed that with the less concentrated DHB matrix solution, the
crystal layer that formed did not completely cover the matrigel surface. This allowed the
incident laser to interact more directly with the matrigel, resulting in the production of
matrigel ions in addition to analyte and DHB matrix ions. With increasing DHB
concentration, the matrigel surface became completely covered with matrix crystals. In
this way, the incident laser only came into direct contact with the matrix crystals
containing the extracted analyte molecules. With the most concentrated DHB solution the
resulting spectra were dominated by (M+H)^ ions and DHB matrix ions.
Optimization of the MALDI Matrix Solvent Polarity
In the next set of experiments the polarity of the matrix solvent mixture was
optimized to see if more of the drug compounds could be made to partition out of the
matrigel tissue and into the MALDI matrix solution. As before, samples of spiperone or
ephedrine in matrigel (100 ppm) were prepared and deposited onto Lasermat sample
33
34
plates for analysis. For these experiments, the DHB matrix solution was prepared at 0.25
M since this gave the best results from the previous experiments. The polarity of the DHB
solvent was varied in each case by using different percentages of acetonitrile and water.
Five solvent mixtures were used ranging from 100% acetonitrile to 100% water. The
polarity of each solvent mixture was calculated using the solvent polarity parameter
defined as:
P'=«iP,+ <^Pb (2-2)
where (f>t and ^ are the volume fractions of water and acetonitrile in the mixture, and P,
and Pb are the polarity values of the pure solvents.*' Polarity values for pure solvents
range from 0.0 - 10.0, v^th zero being the most nonpolar and ten being the most polar; the
polarity of acetonitrile is 5.8 while that for water is 10.0. Table 2-1 shows the polarity
values for the acetonitrile/water solvent mixtures used in this experiment. Solvent
mixtures of acetonitrile and water were used instead of pure solvents because most of the
solvents tested either ran off the surface of the matrigel or dissolved the matrigel tissue
completely. All of the acetonitrile/water mixtures remained as intact droplets on top of
the matrigel samples.
Before analyzing the matrigel samples, MALDI spectra were obtained for standard
solutions of spiperone (2.5x10"^ M) and ephedrine (6x10"^ M) using each of the DHB
solutions prepared in the five different solvent mixtures. The purpose of this experiment
was to determine whether changing the solvent of the matrix solution had a significant
effect on the crystallization process and therefore on the production of the (M+H)"^ ion
35
Table 2-1. Composition and polarity values of the five matrix solvent mixtures used to
optimize the MALDI process for spiperone and ephedrine in matrigel.
Solvent Mixture #
% Acetonitrile
% Water
Polarity Value (P')
1
100
5.8
2
70
30
7.1
3
50
50
8.0
4
30
70
8.9
5
100
10,0
36
signal, independently of the extraction process. Samples were prepared for MALDI by
depositing equal amounts ( 1.0 |iL ) of the standard and matrix solutions on the sample
plate and allowing the mixture to dry and crystallize. The crystallization process for each
of the samples was observed under magnification. With the more volatile solvent mixtures
(>50% acetonitrile) the crystallization process began immediately after adding the matrix
solution. The resulting crystal layer that formed was composed of numerous needles
which covered the entire surface of the sample plate. With the more polar solvent
mixtures, however, the matrix droplet remained for several minutes before crystallizing.
The resulting crystals were noticeably larger but still uniform. The average peak area of
the (M+H)* ion after 100 laser shots for both spiperone and ephedrine using each of the
solvent mixtures is shown in the top plot in Figures 2-7 and 2-8 respectively. The error
bars represent the standard deviation. Mass assignment and peak area values were
generated by the Lasermat software. The Lasermat has a rated mass accuracy of ±0.5 Da
below 5 kDa.*^ Prior to analysis, the instrument was mass calibrated using a standard
solution of leucine enkephalin (M.W. 555) with DHB as the MALDI matrix. As can be
seen from the plots for both standards, changing the solvent mixture did not have a
significant effect on the production of (M+H)* ions.
For spiperone and ephedrine in matrigel, the production of the (M+H)"^ ion signal
improved after increasing the polarity of the matrix solvent mixture (Figure 2-7 and Figure
2-8). One explanation for the trends observed is that adjusting the polarity of the solvent
mixture to match that of the drug compounds achieved a more efficient extraction. This
would explain the fact that ephedrine, a predominantly polar molecule, was more
efficiently extracted from the matrigel using the two most polar solvent mixtures. With
37
^ 1
^ H
z
o
<
a*"
O
lO
I — ^ — I-
O
<
o
I 4-
o
c
o
c
o
0) ^ (0 CO
c2 Co,
85 §E
- CO
- N
- (O
- to
o
o
o
in
CO
s
o
M
<2
o. V
2 >
w o
-a W5
c3 A
IS
.P 3
CO
1
E
+ ^
2 E
4>
O
I
cs
&
3>
U4
4>
O >^
a. J2
38
- O)
O
LO
s
o
CO
r
o
o
o
CM
- 00
- <o
m
9)
e4
T3
C
C
o
cn
X
•c
to
a
c
•c
•o
V
S Q
^ 3
n) b.
+ O
w o
1^
V 00
00 'C
> ^
« c
"5 c •
CI. o
00
I
(S
39
spiperone on the other hand, a gradual increase in the (M+H)^ ion peak area was observed
over the entire range of solvent mixtures because it has both polar and nonpolar parts.
Upon repeating the experiments with matrigel under magnification, however, it was
revealed that using the more polar, less volatile solvent mixtures allowed the matrix
solution to soak on the surface of the matrigel for extended periods of time as was
previously seen with the drug standards. Taking this into consideration, it is more likely
that the trends observed for spiperone and ephedrine in matrigel are a function of both the
extraction efficiency and the soak time of the matrix solution.
Optimization of the MALDI Matrix "Soak Time"
To determine the influence of the soak time of the matrix solution on the
extraction efficiency of spiperone and ephedrine from matrigel a sample procedure was
developed which allowed the crystallization of the MALDI matrix to be more precisely
controlled. Previously, Cottrell and coworkers^* described a procedure for controlling the
crystallization of the matrix solution for peptides bound to nitrocellulose. In their work, a
microscope slide was used to cover the sample on the Lasermat sample plate after the
addition of the MALDI matrix droplet. By allowing the matrix solution to soak for a
couple of minutes the elution of the peptides out of the nitrocellulose and into the matrix
solvent was improved. Using this method, a two-fold increase in analyte ion signal was
reported.
For the experiments with spiperone and ephedrine in matrigel, a Teflon disk
approximately 0.5" in diameter was placed over the raised rim of the Lasermat sample
plate immediately after the addition of the matrix solution. The samples were covered for
40
1 - 30 min. Figure 2-9 shows the plot of the average (M+H)* peak area after 100 laser
shots versus time using the DHB matrix solution prepared in 70% acetonitrile/30% water.
Increasing the soak time from 1-10 min. resulted in the most significant increase in the
(M+H)^ peak area for both compounds. At soak times beyond 15 min., no further
increases were observed. This trend suggests that a partition equilibrium between the
matrigel tissue and the matrix solvent was established. The soak experiments were
repeated with each of the remaining four solvent mixtures. For all of the solvents tested,
increasing the soak time increased the (M+H)^ ion signal of the respective drug
compounds from matrigel. The best results were obtained using the most polar solvent
mixtures at soak times of approximately 5-10 min.
Spatial Resolution Experiments with Spiperone in Matrigel
As outlined at the beginning of the chapter, one of the goals of this work was to
evaluate the potential of MALDI for mapping drug compounds in biological tissues. From
the work presented in the previous sections, it was determined that MALDI increased the
sensitivity for detecting drug compounds in tissue by extracting the analyte molecules out
of the tissue and into the matrix solvent. This has serious implications for applications
involving mapping or imaging of compounds because as the analyte molecules are
extracted out of the tissue, information about their original location in the tissue may be
lost. Theoretically, once the analyte molecules are extracted into the matrix solvent they
are free to migrate in the matrix solution until the solvent evaporates and they become
encapsulated in the matrix crystals.*' The loss of spatial information will be
41
18.
c
2_
a. o>
CO a uj (5
§ E § E
r-
o
s
in
o
CO
to
CM
o
CM
_ o
- II)
0)
E
c
•c
CO
B
|S
o E
0)
•c h
«a s ^
9i
'o.
E 1^
^^^^
CO O
+ u. C
^ — , «
V V >s
at "C
% s
O O
o o
s e
<^
cs
CO
I
42
further compounded by the fact that at longer soak times, analyte molecules can migrate
farther from their initial locations.
To address this issue, spatial resolution experiments were performed with
spiperone in matrigel using the Lasermat instrument. The Lasermat software provides two
aim positions on the sample plate target area 180° apart. The distance between the target
spots was measured experimentally by coating the target area of the sample plate with a
thin layer of Witeout correction fluid. After drying, the sample plate was inserted into the
vacuum chamber of the Lasermat and 100 laser shots were taken at each laser aim
position. Under magnification, two craters measuring approximately 0. 1 mm in diameter
were observed in the surface of the Witeout resulting from laser ablation. The distance
between the centers of the craters at the two positions was measured to be approximately
0.3 mm.
Using a clean sample plate, spiperone mixed with matrigel was added to the left
side of the sample plate (position 1) using a disposable pipet and allowed to dry
completely. A parafilm mask was then laid flat over the entire sample area and cut to
expose only the right side of the sample plate (position 2). This was done to prevent
mixing of the matrigel samples from each position. Matrigel wdthout any analyte was then
deposited onto position 2 (Figure 2-10). After removing the parafilm mask, observation
of the sample plate under magnification showed that two distinct sample regions had been
prepared. To ensure that none of the spiperone from position 1 had migrated to position
2, thirty laser shots were taken at each of the two laser aim positions prior to the addition
of the MALDI matrix solution (Figure 2-11). The spectrum from position 1 was
dominated primarily by low molecular weight matrigel ions along with some spiperone
Figure 2-10. Diagram of the Lasermat sample plate showing the deposition procedure
used for the spatial resolution experiments with spiperone in matrigel.
44
Figure 2-11. Comparison of the LDI spectra obtained from thirty laser shots at
position 1 (top) and position 2 (bottom) on the Lasermat sample plate.
Position 1 was loaded with 100 ppm of spiperone in matrigel. Position 2
contained only matrigel.
45
(M+H)* and (M+Na)"^ ions. No spiperone ions were detected at position 2. For the
MALDI experiment, 1.0 /xL of a 0.25 M DHB solution prepared in 30% acetonitrile/ 70%
water was added over the entire target area of the sample plate. The matrix solution was
allowed to soak on the sample by covering the target area with the teflon disk. After
approximately five minutes the cover was removed and the matrix solution was allowed to
crystallize, covering both positions 1 and 2. Once the crystals had completely dried, the
sample plate was introduced into the ion source of the Lasermat. Thirty laser shots were
again taken at each of the two laser aim positions. With the addition of the DHB matrix,
spiperone was detected at both positions 1 and 2, suggesting that spiperone had in fact
migrated during the soak time of the matrix solution (Figure 2-12). The ratio of spiperone
detected at the two positions was determined fi'om the average (M+H)* peak areas to be
approximately 2:1.
Taking into account the diameter of the target area and the distance of the laser
spot at position 2 fi-om the spiperone-matrigel boundary, the estimated migration distance
for spiperone in matrigel after a five-minute soak time was calculated to be 0.15 - 1.25
mm. These results are significant given the fact that the boundaries between the various
27
regions in complex tissues can be as small as 2 - 5 |im.
An alternative approach to performing MALDI of drug compounds in tissue is to
spray the matrix on top of the tissue surface as a fine mist. The advantage of this
approach is that the migration of the analyte molecules in the tissue can be reduced due to
the rapid evaporation of the solvent and crystallization of the matrix solution. The
electrospray apparatus used in this work consisted of a stainless steel syringe and needle
(flat tip, 18 ga.) mounted vertically above a 0.25" thick stainless steel sample stage.
46
100%= 11250
100
^80
— to
«
30
20-
(M+H)+
m/z 396
DHB Matrix Ions
350 400 460 600 860
200 260 300
DHB Matrix Ions
(M+H)+
m/z 396
200 260 300
3B0 400 480 BOO 650
Mm
(fflW
Figure 2-12. Comparison of the MALDI spectra obtained for spiperone (M.W. 395)
from position 1 (top) and position 2 (bottom) on the Lasermat sample
plate after addition of 1 .0 jiL of 0.25 M DHB matrix solution.
47
The syringe was connected to a 20 kV power supply using a copper lead (Figure 2-13).
The sample stage was connected to the power supply ground. The syringe and sample
stage were housed inside of a Plexiglas chamber for safety concerns. Ceramic rods were
attached to the sample stage to allow the stage to be positioned at variable distances from
the tip of the needle during the spray deposition process.
Using the electrospray setup, the spatial resolution experiments with spiperone in
matrigel were repeated. As before, spiperone mixed with matrigel was applied to the left
side of the sample plate (position 1) while matrigel alone was deposited on the right side
(position 2) using the parafilm mask to prevent mixing. The sample plate was then placed
inside of the electrospray chamber and allowed to rest on the grounded sample stage. The
matrix syringe was filled with 50 ^iL of DHB matrix solution. At the onset of the first
matrix droplet at the tip of the needle, a +5 kV potential was applied to the syringe. The
voltage was increased to +10 kV at which point a fine, uniform, spray was formed which
covered the entire sample plate. At potentials above +10 kV the needle began to vibrate,
causing the matrix spray to miss the sample plate entirely. The distance of the sample
stage from the syringe was also found to have a significant effect on the spray deposition
process. A distance of 1 - 2" below the needle was found to give the most uniform
deposition of the matrix. After about 30 s of spraying, a very thin, homogeneous crystal
layer was formed on the surface of the matrigel sample. MALDI spectra were obtained
for the sample at laser aim positions 1 and 2 as before. By electro spraying the DHB
matrix solution, spiperone was only detected at position 1 (Figure 2-14). The area of the
spiperone (M+H)* peak, however, was only one third of that obtained by applying the
matrix solution as a drop, presumably because spiperone was not as effectively extracted
Figure 2-13. Schematic of the electrospray apparatus used to spray the MALDI matrix
onto the surface of tissue samples.
49
Figure 2-14. Comparison of the MALDI spectra obtained from position 1 (top) and
position 2 (bottom) on the Lasermat sample plate after electrospray
deposition of the DHB matrix solution.
50
from the matrigel. These resuhs suggest that by electrospraying the matrix solution,
migration of the analyte can be minimized when using MALDI. The migration of
spiperone over shorter distances unfortunately could not be investigated due to the fixed
laser aim of the Lasermat instrument.
CHAPTER 3
DESIGN AND CONSTRUCTION OF A NOVEL LASER DESORPTION
QUADRUPOLE ION TRAP MASS SPECTROMETER
After completion of the first MALDI experiments using the Lasermat; research
efforts were focused on designing and constructing a new laser desorption instrument for
MALDI, based on the quadrupole ion trap mass analyzer. While the Lasermat proved to
be usefiil for detecting spiperone and ephedrine in matrigel, it lacked the mass resolution
and MS/MS capabilities needed to detect trace levels of drug compounds fi-om more
complex biological tissues. This chapter introduces the quadrupole ion trap mass analyzer
and provides a detailed description of the various components used to construct the new
instrument. Included is a brief history of the development of the ion trap along with a
description of its function and theory of operation. Also presented in this chapter is a
review of previous laser desorption ion trap designs.
The Quadrupole Ion Trap Mass Spectrometer
Background History
The quadrupole ion trap was first described as a device for storing electrically
charged particles, along with the quadrupole mass filter, by Paul and Steinwedel
(University of Bonn, Germany) in a patent submitted in 1953.*^ In their patent, Paul and
Steinwedel proposed using a combination of radio frequency (RF) and direct current (DC)
51
52
voltages to create a quadrupolar trapping field inside the volume of a solid ion trap
consisting of two endcap electrodes and a central ring electrode (Figure 3-1). Initially, the
ion trap was described as "still another electrode arrangement". However since its
introduction, the ion trap has developed into one of the most sensitive, selective, and
versatile mass spectrometers to date.
Initially, the ion trap was used primarily by physicist to study various physical and
chemical properties of stored ions. In the first work by Paul and coworkers in the early
1950's, the energy absorption of stored ions was measured by applying an RF voltage
across the endcap electrodes of the ion trap.'' The energy absorbed was then related to
the concentration of stored ions. Following this work, Wuerker et al." demonstrated the
storage of small charged particles of aluminum in the ion trap. Experiments were
performed to measure the fi-equency of ion motion by applying a range of supplemental
alternating current (AC) fi-equencies across the endcap electrodes. Photographs were also
taken of the stored particle(s) showing their resonance in the trapping field. Building on
this work, Dehmelt and Major" demonstrated the use of the ion trap for high resolution
spectroscopic studies of ground state, metastable, atomic, and molecular ions.
In 1959, Fischer reported the first use of the ion trap for measuring the molecular
weight of stored ions.'" In his work, Fischer used the ion trap to measure the mass of a
series of krypton isotopes at unit mass resolution using the original mass-selective
detection technique.*' With this technique, the motion of the stored ions was sensed by
means of tuned circuits, such that a response was obtained for each m/z value in turn. The
approach was similar to ion cyclotron resonance in that the ions were detected
nondestructively, inside of the ion trap. In 1968, Dawson and Whetten'^ demonstrated the
June 7, 1960 w. paul etal 2,939.952
A^PWATUS -OR SEPARATINC CHARGED HKTIOXS
OF DirrtRENT SPECIFIC CHARGES
ru*i B.C. 21. 1954 4 Sh..t»^..t <
Figure 3-1 . Sketch of the original quadrupole ion trap taken from the U.S. patent
2,939,952 applied for by Wolfgang Paul and Helmut Steinwedel on
December 24, 1953.*'
54
first use of the ion trap as a true mass spectrometer by using a slightly different approach
called mass-selective storage. In their experiment, different combinations of RF and DC
vohages were applied to the ion trap such that only ions of a single m/z were stable at a
given time. The ions were detected by ejecting them through small holes in one of the
endcap electrodes using a short DC pulse, to an external detector. Satisfactory mass
spectra were obtained using this operational mode, although over a limited mass range.
Almost sixteen years after Dawson and Whetten's work, the first commercially
available ion trap mass spectrometer was introduced by Finnigan MAT in 1984. The Ion
Trap Detector (ITD) 700 was designed as a low-cost benchtop detector for gas
chromatography (GC). With this instrument, ions were formed within the volume of the
ion trap by EI. Ions with m/z values up to 650 could be stored simultaneously inside of
the ion trap by virtue of a 1.1 MHz RF potential applied to the ring electrode. Probably
the most significant development leading to the commercialization of the ion trap was the
mass-selective instability scan developed by Stafford et al. at Finnigan MAT.'* In contrast
to previous detection modes, the mass-selective instability scan involved ramping the RF
amplitude applied to the ring electrode linearly with respect to time. As the RF potential
was increased, ions of increasingly higher m/z developed unstable trajectories inside of the
ion trap and were ejected through holes in one of the endcap electrodes to a detector
(Figure 3-2). Mass spectra were obtained as a function of the RF potential needed to eject
ions of various m/z values to the detector.
Since the introduction of the first commercial ion trap instrument, several advances
in ion trap technology have been made to expand the range of applications of the ion trap
to include the analysis of biomolecules. Of particular note was the advent of resonant
55
Detector
Signal
Isolation
and Storage
kA u.
Mass Analysis
Figure 3-2.
Diagram showing the generation of a mass spectrum using the
the mass-selective instability scan method.
56
ejection or axial modulation in 1988/' The technique of resonant ejection was originally
developed to improve the resolution and dynamic range of the ITD. Variations on this
technique have been developed to allow for the resonant excitation and subsequent
fragmentation of ions inside of the ion trap for MS/MS7* Axial modulation has also been
responsible for extending the mass range of the quadrupole ion trap to well beyond m/z
50,000.™ Advances have also been made in ion isolation with the advent of two-step,*"
apex,** and forward-reverse scan*^ isolation techniques. The application of these
techniques will be discussed in detail in the following sections of this chapter.
The most significant advances in the past seven to eight years have been made in
coupling external ionization sources to the ion trap. These sources allow ions to be formed
outside of the confines of the ion trap. To date, almost every type of ionization source has
been coupled to the ion trap including glow discharge (GD),*^ fast atom bombardment
(FAB),*'* and electron and chemical ionization (EI/CI).*' For the analysis of biomolecules
the two most important ionization methods have been electrospray ionization (ESI)*^ and
MALDI.^"
Ion Trap Theory
The quadrupole ion trap is the three-dimensional analogue of the more common
quadrupole mass filter. However, instead of using four round or hyperbolic rods to create
a quadrupolar trapping field, the ion trap makes use of three symmetrically cylindrical,
hyperbolic electrodes. The central or ring electrode is toroidal in shape and is situated
between two inverted, domed-shaped endcap electrodes (Figure 3-3). The following
57
58
70
derivation was adapted from March and Todd. The general equations defining the
hyperbolic shape of the ring electrode and endcap electrodes are given by:
Mh-2z')=\ (3-1)
•o
A(r^-2z0 = -l (3-2)
2z„
where ro is the inner radius of the ring electrode and zo is the distance from the center of
the ion trap to the endcap electrode. The simplest relationship between the ring electrode
and the endcap electrodes which define a pure quadrupolar field is given by:
ro^ = 2zl (3-3)
A quadrupolar trapping field is generated inside of the ion trap by applying an RF
potential to the ring electrode. The field is uncoupled in the three coordinate directions
(x,y,z). Therefore, the forces acting on an ion are independent of one another and also
vary linearly with the ion's position from the center of the ion trap. The potential applied
to the ring electrode can be represented mathematically by:
<D^=U-Vcosnt (3-4)
where is the applied RF potential, U is the applied DC voltage, V is the zero-to-peak
amplitude of the RF voltage, Q is the angular frequency of the RF trapping field applied to
the ring electrode in rad/s, and t is the time variable. For an ion trap employing an ideal
quadrupolar trapping field, the potential at any given point O, can be represented by:
59
O = -i-—rlr' -2z']+ %° ° ° (3-5)
where is the potential applied to the endcap electrodes. Equation 3-5 is identical to
the general expression for the potential inside of a quadrupole ion trap given by Knight.*'
Substituting equation 3-3 into equation 3-5 and assuming that both endcap electrodes are
grounded (normal operational mode) gives the more common expression:
0 = |^[r^-2z^] + ^ (3-6)
The differential equation of motion for a singly charged positive ion subject to the
potential of equation 3-6 can be obtained from the following:
d^r e
-^ = --VO (3-7)
dt m
where m is the mass of the ion and e is the electronic charge. By inserting equation 3-4
into equation 3-6, setting the field strength at the center of the ion trap to zero (in order to
satisfy LaPlace's equation), and differentiating; the motion of the ion in the radial (r) and
axial (z) directions can be written as:
d^r 2e , .
-1" + r(U-Vcosnt)r = 0 (3-8)
dt 2mro
d^z 4e
+ r(U-Vcosnt)z = 0 (3-9)
dt 2mro
These equations are examples of the Mathieu equations developed 150 years ago to
explain the motion of vibrating membranes.** The general form of the Mathieu equation
can be expressed as:
60
^ + (a„-2q„cos2^)u = 0 (3-10)
where u represents r or z, and |=Qt/2. The stability parameters au and qu determine
whether an ion's motion will be stable or unstable in the quadrupolar field. By performing
a series of operations and substitutions, the stability parameters can be expressed in terms
of the RF and DC potentials applied to the ring electrode and to the m/z of the ion of
interest as follows:
-4eV
Qu =qz =-2q, = / 2 o 2\r^2 (3-12)
m(ro'+2z^)n'
A graphical representation of stable solutions of the Mathieu equation can be generated by
using the dimensionless parameters az and qz as the ordinate and abscissa respectively.
Figure 3-4 shows the regions in (az, qz) space were the radial and axial components of the
ion trajectory are stable. The overlap region near the origin represents the range of az and
qz values that give rise to stable ion trajectories in both the radial and axial directions
simultaneously. This region is more commonly referred to as the Mathieu stability region
(Figure 3-5). When an ion has values of az and qz which fall within the Mathieu stability
region, its motion is stable within the volume of the ion trap.
The lines drawn down and across the stability diagram shown in Figure 3-5 are
called iso-^ lines, and describe the detailed trajectories of ions at that particular point. The
form of an ion trajectory in the r, z plane has the general appearance of a Lissajous curve
composed of two fundamental frequency components o)t,o and <Uz,o of the secular motion.
61
Figure 3-4. Diagram showing the regions of stable ion trajectories in both the radial (r)
and axial (z) directions for the quadrupole ion trap. Regions of simultaneous
overlap are denoted A and B.'°
62
Figure 3-5. Mathieu stability diagram plotted in (a^, qz) space. Ions with &z and Qz values
within this region are stable within the ion trap and can be stored. The lines
running down and across the stability region are iso-P lines used to define the
frequency of oscillation of stored ions.^"
63
with a superimposed micromotion of frequency Q±a) Hz (Figure 3-6). The relationship
between ion frequency and the parameter ^ is given by:
'y„.u=(n + iA)" (3-13)
where 0<^u^ 1 and n = ± 1, ±2. . . . When n = 0, the fundamental frequency of ion motion
reduces to '/2/3ufi.
The values of az and qz as defined by equations 3-11 and 3-12, respectively,
correspond to a single ion isolated in an ideal quadrupole ion trap for which the electrodes
extend to infinity. In order to produce a functional ion trap, however, the electrodes must
be truncated. Truncation of the electrodes introduces higher-order multipole components
to the potential which causes non-linear resonances in ion motion.*' To compensate for
these phenomenon, the ion trap is "stretched" axially by assembling the ion trap in such a
way that the distance between the endcap electrodes is increased by 10.6% (zo = 0.783
cm).^ Despite the stretch in geometry, the stability diagram is changed only slightly.^'
Operation of the Ion Trap
In the normal mode of operation, ions formed inside of the ion trap by EI are
trapped and stored by applying an RF potential to the ring electrode while holding the two
endcap electrodes at ground. The applied RF has a frequency of 1.0485 MHz and an
amplitude of 0-7500 Vo-p. In the instrument constructed for this work, ions were
generated externally by MALDI then directed into the ion trap using a series of focusing
lenses and a DC quadrupole deflector. A -5 V DC offset potential was also applied to the
64
Figure 3-6. Photograph of the ion trajectories for aluminum particles stored inside the ion
trap. The trajectories have a fundamental frequency of motion and a
superimposed micromotion having the form of a Lissajous curve.'^
65
ring and endcap electrodes to control the kinetic energy of the ions as they entered the
trap. Typically, the RF amplitude applied to the ring electrode is set low (qz=0.100 for
miz 100) during the initial ionization event to allow ions over a range of nVz values to be
successfully trapped and stored in the ion trap. For MALDI, however, relatively high RF
amplitudes (qz =0.400) were required due to the high kinetic energy (3-5 eV) of the ions.
A buffer gas of helium (10"^ torr) was also introduced directly into the ion trap. The
purpose of the buffer gas was to increase the trapping efficiency of the ions and to
collisionally cool the ions to the center of the trap where their motion becomes more
stable in the quadrupolar field.
Ion isolation
Ions over a range of m/z values or ions of a single m/z value can be isolated in the
ion trap using a variety of methods including apex,** two-step,*" random noise,'^ stored-
waveform inverse Fourier transform (SWIFT),^^ and filtered noise.'^ The two most
commonly used methods are apex and two-step isolation. In apex isolation (Figure 3-7),
the RF potential applied to the ring electrode is first increased to place the ion of interest
at a qz =0.78 (B). A negative DC potential is then applied to the ring electrode to move
the ion to a point (az=0. 15) just inside the apex of the stability region (C). At the apex, all
ions of m/z greater than the ion of interest are ejected radially from the ion trap, while
those ions with lower m/z values are ejected axially. After isolation, the DC potential is
turned off and the RF amplitude is lowered to position the isolated ion at a more stable qz
=0.30 (A). In two-step isolation (Figure 3-8), ions of m/z greater than that of the selected
ion of interest are ejected fi-om the ion trap across the Pz=0 boundary by applying a
66
-0.7 H 1 1 1 1 1 1 1 1
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
Figure 3-7. Mathieu stability diagram showing the changes in RF and DC voltage levels,
plotted as Qz and az respectively, for apex-isolation. Ions of lower and higher
m/z values are ejected simultaneously at the apex (C).
67
0.2 -I
(0
-0.6 -
Figure 3-8. Mathieu stability diagram showing the changes in RF and DC voltage levels,
plotted as Qz and az respectively, for two-step isolation. Higher m/z ions are
ejected across the Pi=0 line at position (B). Lower m/z ions are ejected
across the line at position (C).
68
positive DC potential to the ring electrode (B). After ejecting the higher masses, the RF
amplitude is increased so that the ion of interest is placed at a q=0.85. A negative DC
potential is then applied to the ring electrode causing all those ions with m/z values lower
than the ion of interest to be ejected across the ^3^=1 boundary (C).
The major drawback of the apex and two-step isolation methods is the fact that
they can only be used to isolate ions up to approximately m/z 600 because of the high DC
voltages required. The molecular weights of the drug compounds studied in this work
ranged fi^om 165 - 1206 Da. In order to isolate the higher mass ions (>m/z 600) an
alternative method known as forward-and-reverse scanning*^ was used (Figure 3-9). In
forward-and-reverse scanning an auxiliary AC vohage (typically 219 kHz, 3-6 Vo-p) is
applied 180° out of phase across the endcap electrodes (A). The main RF is then ramped
up until the frequency of the ion of interest is just below the auxiliary frequency (B).
During the RF ramp (forward scan), ions of m/z lower than the ion of interest sequentially
come into resonance with the supplementary field and are ejected from the trap. After
ejecting the low mass ions, the auxiliary field is turned off" and the RF amplitude is set to
the maximum value (C). The auxiliary field is then turned back on (D) and the RF
amplitude is ramped down (reverse scan); this time until the frequency of the ion of
interest is just above the auxiliary frequency (E). In this way, those ions with m/z values
greater than the ion of interest are resonantly ejected leaving only the ion of interest in the
ion trap (F).
69
Mass of
Resonance c*ou;i;^.
Interest pojnt Stability
Limit
► Forward Scan
(B) — ©ee
(C)
eee
(D)
eee
(E) •QOO'
Reverse Scan
(F)
0 0.908
Figure 3-9. Ion isolation using the forward-and-reverse scan method. Higher and lower
m/z ions are ejected from the ion trap by coming into resonance with a
supplementary AC frequency applied to the endcap electrodes.'"
70
Tandem mass spectrometry (MS/MS)
After isolation, the remaining parent ions can be fragmented inside of the ion trap
to produce daughter ions. These daughter ions, being products of the isolated parent ions
only, can then be used for structure elucidation. This process is known as tandem mass
spectrometry or MS/MS. MS/MS is performed by first setting the RF potential on the
ring electrode to correspond to a qz=0.30 for the parent ion of interest. An auxiliary AC
or "tickle" frequency (6-8 Vo-p) is then applied across the endcap electrodes and tuned to
the frequency of the ions (118 kHz) (Figure 3-10). When the auxiliary frequency matches
the secular frequency of the ions, they become resonantly excited and undergo collisions
with the constant pressure of helium buffer gas in the ion trap. These collisions deposit
energy into the ions which cause them to fragment. This process is termed collision-
induced dissociation (CED).^* By repeating the process of isolating, fragmenting, and
storing the ions; several stages of mass spectrometry or MS" can be carried out inside of
the ion trap.
Ion detection
Ions stored in the ion trap are detected using the mass-selective instability scan'*
where the RF amplitude applied to the ring electrode is increased linearly with respect to
time. Typically, the RF voltage is ramped from 100 - 7500 Vo.^ in 1 10 ms. In this way,
ions of increasingly higher m/z are made to approach the edge of the stability region
(qz=0.908). Once the ions reach the stability edge, they take on axially unstable
trajectories and are ejected from the ion trap to an electron multiplier situated directly
71
72
behind the exit endcap electrode. One of the problems associated with this method is that
during the RF ramp, the higher m/z ions still inside of the ion trap contribute a significant
space charge potential which ultimately degrades the resolution of the resulting mass
spectrum.™ This can be overcome by applying an auxiliary frequency (typically 3-6 Vo-p at
485 kHz) to the endcap electrodes during the analytical scan. As the RP amplitude is
increased, the ions secular motion enters into resonance with the supplementary field
causing the ions to be ejected in a tightly focused packet. This method is known as
resonance ejection or axial modulation.*'
Mass range extension
The maximum m/z ion that can be detected in the normal operational mode of the
ion trap using the mass-selective instability scan with an axially modulation frequency of
485 kHz is 650.™ This upper limit is determined by rearrangement of equation (3-12) to
give:
where V™„=15000 Vp.p, ro= 1 cm, zo= 0.792 cm, QITji = 1.0485 MHz, and qeject= 0.908
are the operating parameters for the Finnigan ITS40 mass spectrometer used in the new
instrument. While the ion trap can be made to trap and store ions of m/z greater than 650;
the RF voltage levels needed to successfiilly eject these ions can not be reached during the
normal mass selective-instability scan. The obvious solution would be to increase the
8eY
max
(m/z)
(3-14)
73
maximum RF voltage applied to the ring electrode. However this would result in severe
arcing between the ring and endcap electrodes.
Several alternative approaches have been proposed to increase the mass range of
the ion trap.'^ In general, these methods involve either reducing the size of the ion trap
(ro, zo), reducing the drive frequency (Q), or lowering the qeject at which ions are ejected
from the ion trap. For all of the high mass analysis performed with the new laser
desorption instrument, the last method was used. In this method, an auxiliary AC
frequency (6-8 Vo^) is applied to the endcap electrodes during the mass-selective
instability scan. However, instead of setting the axial modulation frequency to 485 kHz,
as is normally done for resonance ejection, the frequency is lowered so that ions are made
to come into resonance with the auxiliary field eariier in the RF ramp. As a resuU, the ions
are ejected from the ion trap at lower qeject values (Figure 3-11). By ramping the RF
voltage to its maximum value of 7500 Vo-p, a much higher maximum mass can be ejected
and therefore detected. The plot of the maximum mass detectable for a given axial
modulation frequency is shown in Figure 3-12. One of the consequences of ejecting ions
at lower qeject values, however, is that the assigned mass is lower than the actual mass of
the ion. This is because the ion is ejected at a much lower RF voltage level. To
compensate for this, the assigned mass must be corrected using the following equation: ™
(m/z)„^ =(m/z)„,d*
fa ^
H eject new
V Qeject old J
(3-15)
74
0.2 -1
(0
-0.1 -
-0.2 -
1 L_
ji
Mass
Spectrum
m/z
40
m/z
198
m/z
395
Figure 3-11. Mathieu stability diagram showing resonance ejection of spiperone at three
different axial modulation frequencies: (A) qeject=0.906, (B) qeject=0.454,
and (C) qeject=0.0908. Notice that as the qeject is lowered, the mass
assignment for spiperone is also shifted lower.
75
6500 -
6000 —
5600 —
5000 —
4500 —
(0 4000 -
1 3500 —
CO 3000
2500 —
2000 —
1500 —
1000 -
500 -
lOx
100 200 300 400
Axial Modulation Frequency (kHz)
500
Figure 3-12. Plot of the maximum mass detectable using the mass-selective instability
scan for a given axial modulation frequency.
76
Coupling LDI to the Ion Trap
MALDI Inside the Ion Trap
The early interest in coupling laser desorption ion sources to the ion trap was
driven by the need for improved resolution and by the potential of sequencing high mass
peptides, proteins, and oligonucleotides using the MS" capabilities of the ion trap. Laser
desorption on the quadrupole ion trap was first reported by Cotter and coworkers in
2939 96 ^j^g work, the vacuum housing of a Finnigan ITD 700 was modified to allow
for the insertion of a stainless steel sample probe (Figure 3-13). A flange fitted with a
ZnSe lens was added to the chamber for the introduction of the desorption laser beam.
Modifications were also made to the ion trap as well. Two 0.15" diameter holes were
drilled in opposite sides of the ring electrode. Using this configuration, laser desorbed
ions were produced inside of the ion trap volume for several biomolecules including
sucrose and leucine enkephalin. Fragment spectra were also obtained by gating a beam of
energetic electrons into the trap during the laser desorption event. Building on this work,
Glish et al.'^ demonstrated MS/MS of internally generated laser desorbed ions by CID.
In 1992, Vargas'* used a modified Finnigan ITMS instrument to study the phase
dependency of laser desorbed ions generated inside of the ion trap. In this work, a
stainless steel probe fitted with a graphite tip was positioned at the ring electrode surface.
An RF phase synchronous triggering circuit was then used to trigger the desorption laser
(N2, 337 nm) at phase delay increments of 20°. Using this setup, the trapping efficiency of
C3* graphite ions was found to follow a cyclical pattern comparable to the RF sine wave.
77
78
The first report of MALDI inside of the ion trap was made by Cotter using the
modified ITD instrument used for the first LDI experiments.^ For MALDI, a sample
mixture of matrix and analyte was deposited onto the tip of a 0.09" diameter probe and
inserted into one of the holes in the ring electrode (Figure 3-14). The sample surface was
positioned flush with the inner surface of the ring electrode. To prevent shorting of the
ring electrode, the probe tip was connected to the grounded probe shaft through a Teflon
spacer. Light fi^om a Q-switched Nd:YAG laser (266 nm) was then focused through the
second hole in the ring electrode and made to strike the sample surface using a series of
externally mounted UV quartz lenses. The laser was triggered using the electron gate-
pulse used in the normal operational mode of the ITD to gate electrons into the ion trap
for electron ionization. A second trigger pulse, derived fi^om the RF synchronization
pulses available on the ITD RF board, was input into a delay pulse generator to allow
firing of the laser at preset phase angles of the RF voltage applied to the ring electrode.
Control of the system was provided by the Finnigan ITD software. Using this instrument,
MALDI spectra were obtained in the extended mass range mode (resonant ejection at
lower qeject) for several biomolecules including angiotensin I (M.W. 1296), a- endorphin
(M.W. 1746), and parathyroid hormone (M.W. 3286).
Vargas also reported MALDI inside of the ion trap for several biomolecules
including the drug compound spiperone.''**' MALDI was performed by depositing 2.0 |jL
of a 0.1 ng/piL spiperone solution onto a probe tip followed by an equal volume of 100
mM nicotinic acid matrix solution. MALDI MS and MS/MS spectra were acquired for
spiperone and compared with the corresponding spectra obtained by LDI. Samples of
79
80
spiperone in matrigel were also analyzed by MALDI using DHB matrix with 0.1%
trifluoroacetic acid (TFA).
One of the major drawbacks to performing MALDI inside of the ion trap is that
there is an upper mass limit determined by the kinetic energy of the ions. Beavis and
Chait'*" showed that for MALDI generated ions, the initial kinetic energy increases
linearly with mass (Figure 3-15). Therefore for high mass MALDI ions, the initial kinetic
energy of the ions becomes greater than the pseudopotential well-depth of the quadrupolar
trapping field. Glish calculated a high mass limit of m/z 9830 for MALDI inside of the ion
trap. '"^ Above this limit, the high kinetic energy of the ions causes them to shoot across
the internal volume of the ion trap and strike the opposite side of the ring electrode
without being trapped. Another problem associated with using the internal MALDI
configuration is that late desorbing neutrals fi^om the sample can undergo ion-molecule
reactions with MALDI ions already stored in the ion trap.*^ In performing LDI/MS/MS of
trimethylphenylammonium chloride, Glish et al. observed desorption of neutrals tens of
milliseconds after the initial laser pulse.^' Ion-molecule reactions can interfere with the
MS/MS process, degrade resolution, and can complicate the resulting mass spectrum with
extraneous peaks.
The most significant limitation of the internal configuration for analyzing intact
biological tissues, however, is the fact that the sample is completely enclosed inside the
volume of the ion trap. This is particularly important because in order to perform
microscopy, the microscope objective needed to focus the laser and to view the sample
under magnification must be placed close to the sample surface. For imaging drug
compounds in tissue it is also important that the sample be able to move fi^eely to allow
81
m = 1030
m = 15590
o-*— I — I — I — I I I I I
Energy (cV)
Figure 3-15. Initial translational kinetic energy distributions for a series of polypeptide
ions formed by MALDI.'°'
82
specific regions of the tissue to be targeted for analysis. Neither requirement can be easily
met with the sample positioned inside the ion trap volume.
MALDI Using an External Source Configuration
The first laser desorption work utilizing an external ion source was reported by
Louris and coworkers in 1990.'°^ The setup employed a fiber optic to introduce laser light
into an external ion source located just outside of the entrance endcap electrode. Metal
ions were produced externally by laser desorption then injected with an Einzel lens into the
ion trap where they were allowed to react with neutral benzene molecules.
Following this preliminary work, Bier et al.^" at Finnigan developed the first
MALDI ion trap instrument utilizing an external ion source (Figure 3-16). The instrument
was constructed fi"om an existing commercial Finnigan MAT TSQ 700 triple quadrupole
mass spectrometer. The ion trap was placed in the differentially pumped analyzer region
of the vacuum manifold, replacing the three sets of quadrupole rods. A 3 kV lens was
positioned behind the exit endcap electrode and used to focus ions to a 20 kV conversion
dynode/electron multiplier assembly. The ion source used was the standard TSQ 700
EI/CI source, modified for MALDI by drilling the aperture of the first two extraction
lenses to a diameter of 0.150" to afford a wider angle of acceptance into the optical path.
Two additional holes were drilled through these lenses to allow light fi-om a nitrogen laser
to be introduced into the source through a 200 fim core fiber optic. The fiber optic was
positioned so that the transmitted radiation impinged upon the sample at a 45° angle.
Observation of the sample was made possible by mounting a vacuum flange fitted with a
magnifying glass to the outside of the vacuum chamber just above the ion source. The
83
a
V u
P
I IS
S
8
•a
a
o
Bo
5>
c
SP
•S H o
c c
g .9 a
1 ^ ^
2 5) c
c >n
S u 2
h 60
2 =
E E
(4-1
o
.2 §Q
•g r= u
3 (2
I
u
84
laser was triggered using the normal ionization gate trigger pulse. Additional circuitry was
used to phase-lock the triggering of the laser with the RF vohage.
For MALDI, samples were deposited onto a stainless steel probe and positioned
approximately 0.100" from the first extraction lens in the ion source. The sample probe
was electrically isolated from the ion source block with a Vespel sleeve. Ions desorbed
from the sample surface were extracted by the three-element lens system and transmitted
axially into the ion trap. The lens system and ion trap were operated at relatively high
voltages: - 159 V, - 188 V, -505 V, and -15 V respectively. A small tube lens positioned
inside of the entrance endcap electrode was used to gate ions into the ion trap. The lens
potential was varied between +36 V (gate closed) and -186 V (gate open). Using this
design, intact (M+H)"" ions were obtained for several peptides and proteins up to m/z
43,300 (egg albumin). Results of this work also showed that higher RF vohage levels
were needed during ion injection to trap the higher mass MALDI ions.
Using an external source configuration similar to Bier's original design, Vargas
constructed the first generation MALDI ion trap instrument at the University of Florida.^^
In this instrument, the ion trap was situated in the differentially pumped analyzer region of
a cradle-type vacuum chamber fitted with a quart window flange to allow for the
introduction of the desorption laser beam. The EI/CI ion source used in the instrument
was adapted from a Finnigan 4500 single quadrupole GC/MS instrument.
Samples for MALDI were introduced into the ion source on a stainless steel probe
having a 45° angle tip. Light from a nitrogen laser (337 nm) was focused into the ion
source through a 0. 100" diameter hole in the side of the source block (Figure 3-17). The
hole was originally installed as a GC transfer port. Using this design, Vargas obtained
85
Probe
Sample
Tip
Ion volume
hi)
X
0
D
Ion axis
to ion trap
Focussing lenses
Figure 3-17. External ion source configuration used in the first generation MALDI ion
trap instrument developed at the University of Florida. Samples were
introduced into the ion source on a sample probe having a 45° angle tip."
86
MALDI MS and MS/MS spectra for spiperone using DHB."*^ MALDI was also
attempted for spiperone mixed with matrigel (Figure 3-18). Peaks were observed for
protonated spiperone at m/z 396 and for the sodium adduct at m/z 418. MS/MS could not
be performed for spiperone from matrigel, however, due to the low parent ion signal
intensity. The poor ion signal was believed to result from having the sample situated at a
45° angle with respect to the ion source extraction lenses.
The second generation MALDI ion trap instrument constructed at the University
of Florida by Booth" utilized an external ion source from a commercial Vestec MALDI-
TOF instrument (Figure 3-19). The ion source incorporated three high voltage focusing
lenses (rated at ±35 kV) designed specifically for performing MALDI of high mass
biomolecules. The open configuration of the lens system also allowed samples to be easily
viewed through a large quart wandow mounted above the source region. To maximize the
ion transmission efficiency, both the ion source and ion trap were situated in the source
region of the Vestec vacuum chamber. For MALDI, light from a pulsed nitrogen laser
was focused through a small hole in the second focusing lens and directed onto samples
deposited onto the tip of a stainless steel probe inserted into the face of the first extraction
lens. Using this design, MALDI spectra were obtained for several peptides up to M.W.
2847.5 (melittin). Sequence information was also obtained for the octapeptide
angeotensin II (M.W. 1046.2) by performing MS/MS with helium buffer gas.
The instrument constructed by Booth was also used to analyze tissue samples
prepared for the work presented in this dissertation. Of particular note was the MS*
analysis of spiperone in matrigel and the MS/MS analysis of the anticancer drug taxol
(M.W. 853) from rat liver tissue. A more detailed discussion of the pharmaceutical
87
89
dmg compounds studied will be given in chapter 4. Liver tissue was obtained from a male
Spraige-Dawley rat and immediately frozen for storage. A thin section approximately 2.0
mg in weight was cut and incubated in a 1.0 mL solution containing 100 ng of taxol
(Bristol-Myers Squibb). After a 1 hr incubation period, the section was removed from
solution, washed with deionized water, and placed onto the tip of the MALDI probe. One
microliter of DHB matrix solution was then pipetted on top of the dried liver section and
allowed to dry and crystallize. The MALDI MS and MS/MS spectra for taxol from the
incubated liver tissue is shown in Figure 3-20. With only one stage of mass spectrometry,
the peaks in the spectrum corresponding to taxol were not distinguishable from the intense
background signal from the liver tissue. Using the MS/MS capabilities of the ion trap, the
taxol (M+K)* ion at m/z 892 was resonantly excited and fragmented (after mass isolating
the region between m/z 882 - 902), producing the characteristic taxol daughter ions at m/z
(509+K)^, (569+K)*, and (794+K)^ Comparing the MS/MS spectrum from Figure 3-20
with the daughter ion spectrum from a taxol standard, the presence of taxol in the
incubated liver tissue was confirmed (Figure 3-21). The amount of taxol detected was
calculated to be approximately 360 fg based on the amount of taxol absorbed during the
incubation period and the amount of tissue ablated by the laser.
While the second generation MALDI instrument was capable of detecting drug
compounds from tissues, the limited space in the source region made it impossible to
implement a microscopy system for imaging experiments. The open configuration of the
ion source lens system also ruled out the possibility of performing laser
desorption/chemical ionization (LD/CI) of drug compounds in tissue.
90
Figure 3-20. MALDI MS (top) and MS/MS (bottom) spectra for taxol from rat liver
tissue incubated in a taxol solution (100 ng) for 1 hr.^
91
250 -1
200 -
>, 150 -I
c
0)
- 100 H
50 -
MS/MS of [M + K] of Taxol Standard
[569 + K]
iiii>«)J»iiii>Uii»Dii''ii'i»mi
[509 + K]
300 400
' T '-
500
f
600
m/z
[794 + K]
711
[M + K]
892
/
r
700
JL.
800
900
60
50 -
40 -
V)
20 -
10 -
MS/MS of [M + K] of Taxol in Liver Tissue
[569 + K]
[M + K]
892
300
1 \ 1 r
800 900
Figure 3-21 . Comparison of the MALDI MS/MS spectra for a taxol standard (top) and
for taxol from incubated rat liver tissue (bottom).**
Instrument Design
92
In contrast to the MALDI ion trap instruments discussed so far, the instrument
constructed for this work was designed specifically for the analysis of drug compounds
fi-om intact biological tissues. The instrument consisted of a Finnigan 4500 EI/CI ion
source and a Finnigan ITS40 ion trap mass analyzer housed inside of a differentially
pumped, cradle-type vacuum chamber (Figure 3-22). The ion source was situated 90° off-
axis with respect to the ion trap to allow the desorption laser beam to be directed onto the
sample surface at a 90° angle. With this configuration, either the sample probe or the laser
beam can be manipulated for applications involving imaging of drug compounds in tissue.
The off-axis design was also incorporated to make it easier to position a microscope
objective close to the sample without interfering with the ion trap.
In order to transmit ions formed in the ion source 90° into the ion trap for mass
analysis, a DC quadrupole deflector was used. The ion source and the ion trap were
mounted directly to the DC quadrupole deflector assembly to increase ion transmission
efficiency. After mass analysis, the ions were detected using an electron multiplier
positioned behind the exit endcap electrode of the ion trap. The following sections discuss
in detail the design considerations for the various parts of the instrument.
Vacuum Manifold and Pumping System
The vacuum manifold used to house the working components of the instrument
(ion source, ion trap, DC quadrupole deflector, and detector) was obtained fi^om Finnigan
MAT (San Jose, CA). The design of the manifold was cradle-type and measured 30.0" x
93
94
10.0" X 10.0". The walls of the manifold were constructed of 0.5" stainless steel plate.
Ten 4.0" diameter holes were machined in the walls and base of the manifold to serve as
connection ports for the source, Rf feedthrough, ion gauge, laser window, turbo pumps,
and various other electrical feedthroughs. All port connections were made to fit 5.25" O-
ring vacuum flanges. The top of the manifold was fitted with a 0. 125" 0-ring groove and
sealed with two 1.0" thick glass plates. Inside of the manifold a stainless steel optical rail
was positioned along the floor of the chamber to support the DC quadrupole deflector
assembly and the multiplier. Aluminum brackets were also mounted to the base of the
chamber to support a series of quartz heaters used to heat the vacuum chamber. The
quartz heaters were powered with an external Variac power controller. To allow for
differential pumping, a baffle wall was constructed out of 0.258" thick aluminum plate.
The vacuum manifold was supported on a table constructed from two Finnigan 4500
fi^ames. The original table tops were replaced with two 1.65" thick aluminum plates
measuring 25.5" x 22.0". Two square slots were cut into each plate to accommodate the
turbomolecular pumps.
The pumping system for the vacuum chamber consisted of a TPH 270 L/s
turbomolecular pump (source region) and a TPH 330 L/s turbomolecular pump (analyzer
region), both fi-om Balzers (Hudson, NH). The turbo pumps were mounted directly to the
bottom of the manifold through two 4.0" connection ports. Power was supplied
independently to each of the turbo pumps by two Balzers TCP 300 turbo controllers.
Both turbo pumps were backed by a single 300 L/min mechanical pump (Alcatel
Corporation, Hingham, MA). The pressure in the source and analyzer regions was
monitored using separate Bayard-Alpert type ion gauges (Granville-Phillips, Boulder,
95
CO). The ion gauge used to monitor the source pressure was connected to a Granville-
Phillips model 280 gauge controller equipped with digital readout. The analyzer ion gauge
was connected directly to a Finnigan 4500 vacuum control module. The vacuum
controller was also used to distribute power to both turbo pumps and the mechanical
pump. A vacuum protect mode was supplied with the vacuum controller which cut power
to the pumping system when the base pressure of the manifold exceeded 10"* torr. Using
this pumping system a working base pressure of 1.2(10)"' torr was achieved after
approximately three days.
Ion Source
The Finnigan 4500 EI/CI ion source consisted of a stainless steel source block
containing three electrostatic lenses: two flat, stainless steel focusing lenses and a third
exit tube lens 0.270" in length. The ion source was fitted with a removable, high pressure
ion volume for performing CI. Situated normal to the lens stack was a rhenium filament
and collector cup used to produce the electron beam for EI/CI. Also located in the source
block were four cartridge heaters. The 4500 source was originally designed as an EI/CI
source for gas chromatography. Normally, the GC transfer line was inserted into a small
hole in the side of the source block. Another hole was provided just below the GC port
hole to introduce a calibration gas into the source. Both port holes lead directly to the
central ionization region inside of the source block. In order to perform LD/CI, the
calibration gas line was replaced with a methane reagent gas line. To prevent methane
fi-om leaking out of the source, the GC port was tapped and sealed with a small flathead
screw. A new calibration gas line was positioned in the source block opposite the
96
methane gas line. No other modifications were made to the source block in order to
perform MALDI.
The source block was supported by four 4.0" rods mounted to the inside surface
of a 6.0" Conflat vacuum flange (Figure 3-23). In addition to providing support for the
source block, the source flange was also fitted with a solids probe lock for introducing the
sample probe. The flange was equipped with feedthroughs for the lens system, filament,
collector, and source heaters. Power and control of the filament and the lens voltages was
provided by a Finnigan 4500 lens control module. The voltage range for each of the
lenses is listed in Table 3-1. A gate circuit was used to vary the potential of the exit tube
lens between +170 V (gate closed) and a typical setting of - 24 V (gate open).
One of the reasons for the poor ion transmission efficiency in the first generation
MALDI ion trap instrument was the fact that the ion source block was not directly
mounted to the ion trap assembly. After repeated insertion of the sample probe into the
ion source, the source block eventually became misaligned with the entrance endcap
electrode. To stabilize the ion source in the new instrument, a source adapter was
designed and mounted directly to the two entrance rods of the DC quadrupole deflector
(Figure 3-24). The adapter was machined out of high temperature Macor ceramic to
electrically isolate the source fi"om the rods of the DC deflector. The adapter was
machined to lock into the end of the source block upon insertion of the source assembly
into the vacuum chamber. A 0.5" hole was machined through the center of the adapter to
allow the exit tube lens to pass through easily.
98
Table 3-1. Ion source lens voltages supplied by the Finnigan 4500 lens control module
operated in the positive ion mode.
Ion Source Parameter
Value
Extraction Lens (LI)
-1 to -45 V
Focusing Lens (L2)
-1 to -95 V
Tube Lens (Quad Entrance)
-5 to -25 V
99
0.500'
1.325'
1.151'
— —0.270'
— —0.120'
Material: Macor Ceranic
Figure 3-24. Ceramic adapter for interfacing the ion source block to the DC quadrupole
deflector assembly.
100
DC Quadrupole Deflector Assembly
The use of a DC quadrupole as a 90° deflector was first described by Zeman for
use with a laser-ion coaxial beam spectrometer.'*^ In 1988, Russell et al.'**^ reported a
preliminary design for a time-of-flight Fourier transform mass spectrometer (TOF-FTMS)
employing a DC quadrupole deflector for beam steering. Pedder and Yost in 1989
described the use of a DC quadrupole deflector to transmit ions formed by EI/CI into the
body of a quadrupole ion trap.'"* For the new laser desorption instrument, the DC
quadrupole deflector was chosen over other varieties of beam steering devices because it
did not distort the ion beam appreciably after it had been turned 90° and because the gap
between the quadrupole rods allowed the desorption laser beam to pass through easily.
The DC quadrupole deflector was constructed fi-om four quarter round, stainless steel
rods of radius 0.50" and length 3.25". For added stability, the rods were bolted to the
inside surface of two anodized aluminum caps. The rods were secured to the caps with
Teflon screws to prevent grounding. Opposing rods were electrically connected using
copper wire. A stainless steel tube lens was also mounted to the two quadrupole rods
situated normal to the ion source adapter to help focus the ions into the ion trap. Special
washers and sleeves were machined out of Torlon, a high temperature ceramic, to provide
support and electrical isolation for the tube lens. To assure proper alignment of the tube
lens with the ion trap the end of the tube lens was machined to fit inside the entrance
endcap electrode of the ion trap. The ion trap was held in place using a mounting plate
supported by two aluminum rods attached to the anodized aluminum mounting blocks of
the DC quadrupole assembly (Figure 3-25). The entire DC quadrupole deflector assembly
101
CD
4-> Qj
c
o
u
d
L
pq
Qj
O
a
d
G5
U Qj
o a
o
-p
u
Hi
o £
o «
— Q.
J3 CO
S is
o 1
^ o
u
T3
o
CL
e «
3
W
Q "
5
3
O
ai
e
o
2 «
g o
V o.
0 T3
1 s
o
u
ca
£
(N
I
3)
102
including the attached ion trap was then mounted to an aluminum support plate which
allowed the assembly to be vertically aligned with the ion source. Power and control of
the DC quadrupole deflector assembly was provided by a Finnigan 4500 Quadrupole
Electronics Module (QEM). The voltages supplied by the QEM are listed in Table 3-2.
The DC quadrupole was operated by aligning the quadrupole rods vertically and
applying different combinations of voltages to the two sets of opposing rods. In this way,
a quadrupolar field was defined in the x,y plane, perpendicular to the rods. Ions entering
between the first two adjacent rods were turned 90° around one of the quadrupole rods
and focused into the ion trap through the tube lens. One of the drawbacks of the DC
quadrupole deflector is that there is no focusing of the ion packet in the z-direction
(parallel to the rods).''*' To compensate for this, relatively high vohages were applied to
the tube lens in order to refocus the ion beam after it had been turned 90°. In addition to
fiinctioning as a beam steering device, the quadrupole deflector also acts as an energy
analyzer. In previous work, Pedder'"' showed by simulation that ions of virtually any
energy could be transmitted with 60-70% efficiency by properly adjusting the voltages
applied to the quadrupole rods. However, for a given set of voltages the energy window
was found to be rather narrow. A fiill characterization of the DC quadrupole deflector has
been given by Zeman and more recently by Pedder.
Laser Setup
The laser employed in the new instrument was a Laser Science Inc. (Cambridge,
MA) model VSL-337ND pulsed nitrogen laser operating at a wavelength of 337.1 nm
with a spectral bandwidth of 0.2 nm. The laser had a 3 ns pulse width (FWHM) with an
103
Table 3-2. DC quadrupole assembly voltages supplied by the Finnigan 4500 QEM.
Instrument Parameter
Value
Quad Pair 1
Oto-10 V
Quad Pair 2 (Turning Quads)
0 to -97 V
Tube Lens
Oto-130 V
Trap Offset
+30 to -30 V
104
average jitter of ±10 ns, and a repetition rate <20 Hz. The laser energy was >250
^J/pulse with a peak power of 85 kW. Pulse-to-pulse stability was given as ±4% at 10
Hz repetition rate. The laser was near-diffraction limited which allowed the laser beam to
be more easily focused to a small spot. The beam cross section was 40 mm^. The laser
setup is shown in Figure 3-26. The laser was placed on an aluminum table taken from the
Vestec MALDI-TOF instrument and positioned parallel to the vacuum chamber. The
laser table was fitted with adjustable feet to allow the height of the laser to be varied. The
laser beam was deflected using a right angle prism and directed through a low distortion
quart window mounted in the side of the vacuum chamber opposite the ion source.
Before entering the vacuum manifold the laser beam was focused down to a spot size of
approximately 0.13 mm^ (as measured on the sample surface under magnification) using a
single focusing lens (Melles-Griot). The focal length of the lens was 25 .4 cm. The beam
intensity was adjusted between 10* -lO' W/cm^ using a wheel attenuator (Newport Corp.)
situated between the laser and the prism. Alignment of the laser was accomplished by
observing the fluorescence from a drop of Witeout correction fluid deposited and dried on
the sample probe tip. The position of the laser spot on the sample probe was controlled
manually using an x,y,z-niicromanipultor attached to the prism (Newport Corp.). Rough
positioning was also accomplished by simply rotating the sample probe. The laser was
triggered externally using the rising edge of a TTL pulse provided by the Finnigan ITS40
electronics. A 3 ms delay was added to the laser trigger pulse using a Wavetek model 275
function generator (Indianapolis, IN). This was necessary to ensure that ions were being
formed after the tube lens in the ion source had been gated open. No attempt was made to
phase-lock the triggering of the laser with the main RF applied to the ring electrode.
106
Software Control
The electronics used to control the operation of the ion trap were obtained fi-om a
Finnigan ITS40 GC/MS instrument. The ITS40 electronics module has a 80186
microprocessor located on the scan and acquisition (SAP) board which controls the RF
amplitude and frequency and the DC voltages applied to the ring and endcap electrodes
during all stages of ion trap operation (Figure 3-27). In addition, the SAP also processes
data acquired fi"om the detector and downloads it to the PC where it can be displayed as a
mass spectrum. The software used to control the SAP (Gatorware) was written by Tim
Griffin and Nathan Yates at the University of Florida."**'"' Gatorware allows the user to
control the operation of the ion trap through lists of instructions known as scan tables
(Figure 3-28). Each scan table tells the SAP what voltages to apply to the ion trap for a
specified amount of time. The Gatorware software also provided computer control over
the auxiliary board used to set the axial modulation frequency and amplitude for
performing mass range extension and CID.
107
Amplifier and
RF Generator,
Fundamental
RF Voltage
Scan Acquisition
Processor
(SAP)
A
V
Amplifier and
RF Generator,
Supplementary
RF Voltage
|xjla-X|
Central Processing Unit
(CPU)
Hard Drive
1
Figure 3-27. Block diagram depicting the interconnections between the ion trap, the
SAP board of the ITS40 electronics module, and a personal computer.
Scan Table Editor
Sciin Ion Trap
100.0
0.01260
AGC Editor
Aux Editor
SAP Table
CALC SCAN-»SAP
Table 12
Table 14
0 +1 +10 +100
= -1 -10 -100
Figure 3-28. Example of a scan table provided by the Gatorware software.
CHAPTER 4
MALDI OF DRUG COMPOUNDS IN TISSUE USING A QUADRUPOLE ION TRAP
MASS SPECTROMETER
In the final stage of this project, applications of the new laser desorption
instrument were made for the analysis of three significant pharmaceutical drug compounds
spiperone, taxol, and polymyxin Bi. The goal of this work was to demonstrate the
usefiilness of MALDI coupled with the MS/MS capabilities of the ion trap for detecting
trace levels of these drug compounds from complex biological tissues. This chapter
begins with a brief review of the initial experiments performed to tune and calibrate the
new instrument. Following this section are the experimental results for spiperone in rat
cerebral tissue, taxol in mouse ovarian tumor tissue, and polymyxin Bi in human plasma.
Also included with the results are introductions to each of the drug compounds studied,
along with a detailed description of their pharmacological use and mechanism of action.
The chapter concludes with the results for the initial laser desorption/chemical ionization
experiments with trimethylphenylammonium bromide.
Instrument Calibration & Optimization
EI of Perfluorotributylamine Calibration Gas
Before performing MALDI of drug compounds in tissue, the new laser desorption
instrument was calibrated using perfluorotributylamine (PFTBA M.W. 670.96), a common
109
110
calibration compound used in many commercial mass spectrometers. PFTBA is
particularly usefijl for mass calibration because it produces fragment ions spanning the
normal mass range of the ion trap. The calibration gas was introduced into the ion source
using the standard 4500 needle valve inlet mounted to the outside face of the ion source
flange. The pressure inside the source region of the vacuum manifold was measured at
6x10"^ torr (uncorrected) using a Bayard- Alpert ion gauge. Helium buffer gas was also
introduced into the analyzer region at an indicated pressure of 8x10"* torr. EI was
performed with a 70 eV electron beam from the rhenium filament situated in the ion
source block. The EI spectrum obtained for PFTBA is shown in Figure 4-1. The
corresponding EI scan function parameters used to acquire this spectrum are listed in
Table 4-1. Mass calibration was performed using the Gatorware software. PFTBA was
also used to tune the voltages for the ion source lenses, DC quadrupole rods, and the ion
trap offset by observing the intensity of the fragment peaks as a function of the various
voltage level settings. The optimized voltages for EI are listed in Table 4-2.
After tuning and mass calibrating the instrument with PFTBA, MALDI was
attempted using a standard solution of spiperone in methanol. A small volume of the
standard was mixed with a solution of DHB matrix and deposited on the tip of the sample
probe. After allowing the sample to dry and crystallize, the sample probe was inserted
into the vacuum manifold through the solids probe lock and positioned in front of the first
extraction lens in the ion source. The EI scan function used to obtain the spectrum for
PFTBA was modified for MALDI by adding six laser trigger tables (Table 4-3). A 50 ms
cool table was also added after each laser trigger table to allow the laser to recharge
between fires. The RF level during injection was increased to q = 0.400 for a table mass
Ill
700 -I
c
600 -
500 -
400 -
300 -
200 -
100 -
131
69
100
/
119
/
264
414
502
614
50 100 150 200 250 300 350 400 450 500 550 600 650
m/z
Figure 4-1. EI spectrum of perfluorotributylamine (PFTBA) calibration gas.
112
Table 4-1 . EI scan function parameters used for the analysis of PFTBA.
Table
(RF
Volt.)
#
Scan Table
Mass
Start q
End q
1
Reset
100
0.000
0.000
2
Preionize
100
0.180
0.180
3
Ionize
100
0.180
0.180
4
Cool
100
0.180
0.180
5
Prescan
100
0.180
0.180
6
Scan
100
0.180
0.890
7
Empty Trap
100
0.000
0.000
(Axial Modulation)
(DC Volt.) Freq. Amp. Time
Start a* End a (kHz) (V) (ms)
0.0189 0.0189 _ _ 1
0.0189 0.0189 _ _ 0.5
0.0189 0.0189 _ _ 30
0.0189 0.0189 _ _ 1
0.0189 0.0189 485 0 0.5
0.0189 0.0189 485 4 110
0.0189 0.0189 1
* corresponds to a trap offset of - 12V.
113
Table 4-2. Optimized instrument parameters for EI.
Instrument Parameter
Voltage Level
Extractor Lens (LI)
-11.7 V
Focusing Lens (LI)
-52.6 V
Tube Lens (Quad Entrance)
-18.2 V
Quad Pair 1
0 V
Quad Pair 2
-53 V
Tube Lens (Trap Entrance)
-64 V
Trap Offset
-12 V
114
Table 4-3. MALDI scan function parameters used for the analysis of spiperone.
(Axial Modulation)
Table
(RF
Volt.)
(DC
Volt.)
Freq.
Amp.
Time
#
Scan Table
Mass
Start q
End q
Start a*
End a
(kHz)
m
(ms)
1
Reset
100
0.000
0.000
0.0042
0.0042
—
—
1
2
Preionize
100
0.400
0.400
0.0042
0.0042
—
—
0.5
3
Trig Laser 1
100
0.400
0.400
0.0042
0.0042
—
—
10
4
Cool
100
0.400
0.400
0.0042
0.0042
—
—
50
5
Trig Laser 2
100
0.400
0.400
0.0042
0.0042
-
—
10
6
Cool
100
0.400
0.400
0.0042
0.0042
—
—
50
7
Trig Laser 3
100
0.400
0.400
0.0042
0.0042
-
-
10
8
Cool
100
0.400
0.400
0.0042
0.0042
-
-
50
9
Trig Laser 4
100
0.400
0.400
0.0042
0.0042
—
—
10
10
Cool
100
0.400
0.400
0.0042
0.0042
50
11
Trig Laser 5
100
0.400
0.400
0.0042
0.0042
10
12
Cool
100
0.400
0.400
0.0042
0.0042
50
13
Trig Laser 6
100
0.400
0.400
0.0042
0.0042
10
14
Cool
100
0.400
0.360
0.0042
0.0042
50
14
Prescan
100
0.360
0.360
0.0042
0.0042
485
0
0.5
15
Scan
100
0.360
0.890
0.0042
0.0042
485
4
110
16
Empty Trap
100
0.000
0.000
0.0042
0.0042
1
* corresponds to a trap offset of - 5V.
115
of 100. The voltage settings for the source lenses, the DC quadrupole deflector, and the
ion trap offset were unchanged. No peaks were observed for spiperone or DHB using the
optimized EI voltage settings.
Instrument Simulation using SIMON V6.0s
In order to tune the instrument for MALDI, a computer simulation program,
SIMION V6.0s, was employed. SIMION V6.0s (Idaho National Engineering Laboratory,
ID) is a C-based program designed to model electrostatic ion optics elements via 2D and
3D potential arrays. The potential arrays are sized, oriented, and positioned as instances
within an ion optics workspace. Ions are then allowed to move within the workspace to
determine how the fields generated by the potential arrays impact the ions' trajectories.
Visualization features in SIMION V6.0s allow the user to cut into any component to
more closely inspect ion trajectories and potential energy surfaces. Another important
feature of SIMION V6.0s is that it allows the various potentials assigned to each array to
be adjusted quickly and independently of one another.
For the simulation experiments, 2D potential arrays were generated for the ion
source, the DC quadrupole deflector assembly (including the tube lens), and the entrance
endcap electrode of the ion trap (Figure 4-2). The tip of the sample probe was also
included for the MALDI simulations. The entire ion trap was not modeled to save
memory space and to cut down on the time needed to refine the individual arrays. The
grid spacing for the arrays was set to 0. 1 mm/grid. In the first simulation, a packet of ten
ions was started in the ion source with an initial kinetic energy of 0.05 eV (typical K.E.
116
Ion Trap
Figure 4-2. SIMION V6.0s potential arrays for the ion source lenses (Ext, Lens, Quad
Ent), DC quadrupole deflector (Quad 1, Quad 2), tube lens, and the ion trap
entrance endcap electrode of the new laser desorption instrument.
117
for EI generated ions). The m/z of the ions was set to 396 to simulate the trajectories for
spiperone ions. The potentials of the various arrays were set to correspond to the
experimentally optimized EI voltage settings. Using these voltages the ions were
successfully focused into the DC quadrupole deflector, turned 90° around Quad 2, and
injected through the entrance endcap aperture (Figure 4-3).
For the second simulation experiment the same potential arrays and voltage
settings were used, but the initial kinetic energy of the ions was increased to 2.5 eV to
simulate the trajectories for spiperone ions generated by MALDI. Beavis and Chait
determined that MALDI produced ions above m/z 1000 have similar velocity distributions
and travel at an average velocity of 750 m/s.'°' Lower mass ions, however, are not cooled
to the same degree in the expanded supersonic jet, and therefore travel at velocities
approaching 2000 m/s. From interpolation, the velocity of spiperone ions formed by
MALDI was estimated to be approximately 1 150 m/s. The simulated trajectories for the
MALDI-generated spiperone ions using the EI voltage settings is shov^ in Figure 4-4.
With the higher initial kinetic energy, the ions were not sufficiently turned by the
quadrupole rods and ended up hitting the tube lens before making it into the ion trap.
These resuhs explain the inability to acquire MALDI spectra for spiperone experimentally
using the EI voltage settings. To compensate for the increased kinetic energy of the
MALDI ions, the voltages for Quad 2 and the tube lens were increased to -68V and -81V,
respectively. Using these optimized SIMION voltages, the first experimental MALDI
spectrum for spiperone was obtained (Figure 4-5).
118
m Quad 1:0V
Ion Mass: 396 Da Ext: -11. 7 V Quad 2: -53 V
Kinetic Energy: 0.05 eV Lens: -52.6 V Tube Lens: -64 V
Number of Ions; 10 Quad Ent: -18.2 V Trap Offset: -12 V
Figure 4-3
Simulated trajectories for EI generated spiperone ions (0.05 eV) using the
experimentally optimized EI voltage settings.
119
MALDI Quad 1:0 V
Ion Mass: 396 Da Ext: -11.7 V Quad 2: -53 V
Kinetic Energy: 2.5 eV Lens: -52.6 V Tube Lens: -64 V
Number of Ions: 10 Quad Ent: -18.2 V Trap Offset: -12 V
Figure 4-4. Simulated trajectories for MALDI generated spiperone ions (2.5 eV) using the
experimentally optimized EI voltage settings.
120
500 -1
400
300 -
c
200 -
100 -
137
274
(M+H)
396
i Lilt J
50 100 150 200 250 300 350 400 450
m/z
500
Figure 4-5. MALDI MS spectnim of spiperone (M.W. 395) using DHB matrix.
121
The dependence of the voltage settings for the quadrupole rods and the tube lens
on the initial kinetic energy of the ions was further investigated by simulating the
trajectories for ions of increasing m/z up to m/z 3000 (10 eV). Although the average
velocity of MALDI ions above m/z 1000 remains constant at 750 m/s, their kinetic energy
increases linearly with mass.'"' From the simulations, it was found that as the kinetic
energy of the ions increased, higher voltages were required on Quad 2 and the tube lens to
turn and focus the ions into the ion trap. For ions above m/z 1000, a small positive
potential (2 - 5 V) was also needed on Quad 1 .
High Mass Calibration using a Peptide Mixture
Normally, the ion trap is calibrated using the Gatorware software at a resonant
ejection frequency of 485 kHz (qqect = 0.906). For all of the high mass experiments (>m/z
650) presented in this chapter, the resonant ejection frequency was set to 130 kHz. As
discussed previously in chapter 3, one of the consequences of extending the mass range by
using lower resonant ejection frequencies is that the ions are ejected at lower RF voltages.
Because the Gatorware software sets the mass calibration curve according to a qeject =
0.906, the assigned masses for ions using the lower resonant ejection frequency were
significantly lower. To calibrate the instrument for the extended mass range (~ 2000 Da),
MALDI was performed using a mixture of three peptide standards: methionine-arginine-
phenylalanine-alanine (MRFA M.W. 523.6), methionine enkephalin-arginine-glycine-
leucine (M.W. 900.4), and angiotensin II (M.W. 1046.2). Working standards for these
peptides were prepared at 2x10"' M in deionized water. Equal amounts (50 ^iL) of each
standard solution was then mixed in a 0.5 mL vial. For MALDI, 10 (oL of the peptide
122
mixture was mixed with an equal volume of a 0.5 M DHB matrix solution prepared in
50% acetonitrile/50% water. One microliter of the mixture was deposited on the tip of the
sample probe. Using the optimized SIMION voltages and the MALDI scan function
parameters for spiperone (using an axial modulation frequency of 130 kHz), the MALDI
spectrum shown in Figure 4-6 was obtained. Slightly higher voltages were needed on
Quad 2 and the tube lens to obtain a satisfactory signal for the higher mass angiotensin II
ions. The major peaks in the spectrum at m/z 525, 901, and 1047, corresponded to the
(M+H)"" ions for MRFA, methionine enkephalin-arginine-glycine-leucine, and angiotensin
II, respectively. Significant sodium adduct peaks were also observed for MRFA (m/z 547)
and angiotensin II (m/z 1069). The low intensity peaks below m/z 300 corresponded to
ions from the DHB matrix. The calibration curve for the extended mass range (y =
2.63944x + 1.89841, where y = corrected m/z and x = m/z at 485 kHz) was generated by
plotting the actual mass of the ions versus the observed mass assigned by the Gatorware
software.
Analysis of Spiperone from Rat Cerebral Tissue
Spiperone was developed by Bristol-Myers Squibb as an antipsychotic drug for the
treatment of various neurological diseases including schizophrenia. Currently,
schizophrenia affects 1 in every 100 people worldwide between the ages of 16 and 30."^
Schizophrenia is caused by an imbalance in the levels of neurotransmitters, the substances
that allow communication between nerve cells, in the brain. The symptoms associated
with schizophrenia are varied, but generally include severe thought and speech
disturbances, hallucinations, delusions, anxiety, and uncontrollable behavior.
123
1400 -1
1200 -
1000 -
800 -
c
600 -
400 -
200 -
MRFA
(M+H)
524
Met. Enkephalin-Arg-Glv-Leu
(M+H)
901
Angiotensin II
100 200 300 400 500 600 700 800 900 1000 1100 1200
m/z
Figure 4-6. MALDI MS spectnam of a peptide mixture of MRFA (M.W. 523.6),
methionine enkephalin-argenine-glycine-leucine (M.W. 900.4), and
angiotensin II (M.W. 1046.2).
124
Of particular importance in neurological disorders such as schizophrenia is the role
of the neurotransmitter serotonin (5-hydroxytryptamine). Serotonin, or 5-HT, is
synthesized in brain neurons from the amino acid tryptophan and is stored in vesicles in the
synaptic terminal.*^ Upon a nerve impulse, serotonin is released into the synaptic cleft and
binds to specific receptor sites on the adjacent neuron. Current research has identified at
least four populations of receptors for serotonin, 5-HTi, 5-HT2, S-HTs, and 5-HT4, each
having various subtypes."'* The 5-HTia receptors are located primarily in the central
nervous system and are associated with depression, anxiety, and other psychiatric
disorders. The onset of severe depression occurs when the neural pathways in the brain
are understimulated due to the lack of serotonin release into the synaptic junction.
Conversely, when the release of serotonin is too great, the synaptic junction becomes
flooded, causing overstimulation of the neurons. People with schizophrenia show
unusually high levels of activity in specific regions of the brain during hallucinations and
periods of anxiety."' The actions of serotonin can be modulated by drugs that either
block its storage, stimulate or inhibit its release, or mimic or inhibit its action at various
postsynaptic receptors.
Spiperone belongs to a class of compounds know as azipirones which are similar in
structure to serotonin. These compounds bind to 5-HTia receptors in the central nervous
system and inhibit the firing of the of 5-HT neurons."' In this way, the number of
sympathetic nerve discharges (SNDs) is reduced. The problem with spiperone as an
antipsychotic drug, however, is that it does not selectively bind to 5-HT 1 a receptors.
Instead, spiperone also has high affinity for 5-HT2 and dopomine D2 receptors which can
stimulate SNDs."' Because these receptors are present at different densities throughout
125
the brain, there is great interest in studying the selectivity of spiperone by determining its
concentration in various cerebral regions. As a first step towards potentially mapping
spiperone and other psychotrophic drug compounds in the brain, experiments were
performed on rat cerebral tissue incubated in spiperone.
MALDI MS and MS/MS of Standard Spiperone
In order to confirm the presence of spiperone fi"om rat cerebral tissue, reference
MS and MS/MS spectra were first obtained for a spiperone standard. A stock solution of
spiperone was prepared by dissolving 0.01 g of the solid material in 25 mL of methanol.
A 1:10 dilution of this stock solution was used to make the final 10"* M standard solution.
One microliter of the standard solution was mixed with 1.0 |iL of DHB matrix solution
(0.1 M) on the tip of the sample probe to give a molar ratio of 1:1,000. The MALDI
spectrum acquired for spiperone was dominated primarily by the (M+H)"^ peak at m/z 396
and lower m/z ions corresponding to the DHB matrix.
After obtaining a steady signal for spiperone, MS/MS was performed by first
isolating the (M+H)* ion using the forward-and-reverse scan method. After isolation, the
(M+H)"^ ion was resonantly excited and fi-agmented by CID (115 kHz, 8 V, 20 ms) using a
constant indicated pressure of helium buffer gas in the analyzer region of 1x10'* torr. The
MS/MS spectrum for spiperone, including the proposed fi-agmentation pathways is shown
in Figure 4-7. The major daughter ions observed for spiperone were m/z 290, m/z 265,
m/z 230, and m/z 165. The MALDI MS/MS scan function parameters are listed in Table
4-4.
126
Figure 4-7. MALDI MS/MS spectrum of a spiperone standard.
127
Table 4-4. MALDI MS/MS scan function parameters used for the analysis of spiperone.
Table
Volt.)
#
Scan Table
Mass
Start q
End q
1
Reset
100
0.000
0.000
2
Preionize
100
0.400
0.400
3
Trig Laser 1
100
0.400
0.400
4
Cool
100
0.400
0.400
5
Trig Laser 2
100
0.400
0.400
6
Cool
100
0.400
0.400
7
Trig Laser 3
100
0.400
0.400
8
Cool
100
0.400
0.400
9
Trig Laser 4
100
0.400
0.400
1 \J\J
0 400
0 400
11
Trig Laser 5
100
0.400
0.400
12
Cool
100
0.400
0.400
13
Trig Laser 6
100
0.400
0.400
14
Cool
100
0.400
0.400
15
Eject Low
Masses
396
0.101
0.860
16
Eject High
Masses
396
0.860
0.310
17
Cool
396
0.310
0.300
18
CID
396
0.300
0.300
19
Cool
396
0.300
0.091
20
Prescan
100
0.360
0.360
21
Scan
100
0.360
0.890
22
Empty Trap
100
0.000
0.000
(Axial Modulation)
CDC
Volt
J. 1 •
Amn
Start a
End a
(kHz)
m
fms)
0.0042
0.0042
1
0.0042
0.0042
0.5
0.0042
0.0042
10
0.0042
0.0042
50
0.0042
0.0042
10
0.0042
0.0042
50
0.0042
0.0042
10
0.0042
0.0042
50
0.0042
0.0042
—
—
10
0.0042
0.0042
-
-
50
0.0042
0.0042
10
0.0042
0.0042
50
0.0042
0.0042
10
0.0042
0.0042
50
-.0017
-.0017
485
5
0.5
-.0017
-.0017
115
5
0.5
-.0017
-.0017
1
-.0017
-.0017
115
8
20
-.0017
-.0017
1
0.0042
0.0042
485
0
0.5
0.0042
0.0042
485
4
110
0.0042
0.0042
1
128
Preparation of the Cerebral Tissue
For the tissue experiments with spiperone, whole brain was obtained from a male
Sprague-Dawley rat immediately after the animal was sacrificed. The brain was washed
with several aliquots of HEPES buffer solution and frozen (-20° C) for storage in a small
plastic container. While the brain was still frozen, it was sectioned into two halves
between the right and left parietal lobes. A thin slice approximately 0.5 mm thick was cut
from the inner portion of the left cerebral hemisphere using a disposable scalpel. The
section was cut to include portions of the cerebral cortex, hippocampus, corpus callosum,
occipital cortex, and cerebellum. Before incubation, the tissue was trimmed to fit the
diameter of the sample probe tip (5.0 mm) and weighed on a sheet of weighing paper.
The weight of the tissue section was measured to be 2 .0 mg. After weighing, the cerebral
tissue was transferred to a small glass vial containing 1.0 mL of a 10"^ M solution of
spiperone in methanol. After an incubation period of 1 hr. the section was removed from
the vial, taking special care not to fold or tear the tissue. The tissue was gently shaken
then washed with several drops of water to remove any excess spiperone solution
remaining on the tissue surface. The weight of the tissue after incubation was 9.2 mg.
Under microscopic observation the tissue appeared swollen with narrow, evenly spaced
ridges on the surface. No differentiation between the various regions of the brain was
noted.
129
MALDI Analysis of Cerebral Tissue
For MALDI, the tissue section was first positioned on the tip of the stainless steel
sample probe then flattened using the end of a spatula. 1 - 2 of 0. 1 M DHB matrix
solution was then pipetted onto the surface of the tissue and allowed to soak for several
minutes. After approximately 10 minutes the matrix solution had completely dried and
crystallized, covering the entire surface of the tissue. The addition of the DHB matrix was
also found to help adhere the tissue to the sample probe. For analysis, the sample probe
was inserted into the ion source and the laser was fired at various positions on the sample.
In contrast to the earlier experiments with spiperone in matrigel, the shot-to-shot
reproducibility of the ion signal fi^om the cerebral tissue was rather poor. This was caused
by the formation of several pockets in the tissue surface upon further drying of the sample
in the vacuum chamber. Once a satisfactory signal was obtained for spiperone; MS/MS
was performed using the same scan function parameters used for the spiperone standard.
The MALDI MS/MS spectrum for spiperone from the rat cerebral tissue is compared with
the MS/MS spectrum for the spiperone standard in Figure 4-8. The spectrum was
acquired from the sum of eighteen laser shots at a single spot on the cerebral tissue. After
analysis, the tissue sample was observed under the microscope. Several small laser holes
were observed in the tissue in addition to the pockets mentioned earlier. However, the
exact location of the laser spot corresponding to the MS/MS spectrum shown in Figure 4-
8 could not be determined. Based on the thickness of the tissue section (0.5 mm), the
volume of tissue sampled by the laser (8.5x10"^ mm^), and the calculated amount of
130
500 -1
400 -
300 -
c
- 200
100 -
MS/MS of (M+H) of Spiperone
in Rat Cerebral Tissue
(M+H)
396
— I 1 1 1 1 1 1 1 1 1 1 1 ■ 1 1 1
50 100 150 200 250 300 350 400 450
m/z
1200 -, MS/MS Of (M+H)'*'of a Spiperone Standard
1000 -
800 -
c 600
03
400 -
200 -
165
290
230
50
n r
100 150
265
(M+H)
396
200
I
250
m/z
300
350 400
450
Figure 4-8 Comparison of the MALDI MS/MS spectrum for spiperone from rat
cerebral tissue (top), with the corresponding spectrum for a spiperone
standard (bottom).
131
spiperone solution absorbed by the tissue after incubation (7.2 mg); the amount of
spiperone sampled was determined to be approximately 311 pg.
Analysis of Taxol from Mouse Ovarian Tumor Tissue
The second drug compound studied was the anticancer agent taxol. Taxol is the
trade name for paclitaxel, a member of the Taxus alkaloid family of natural products found
in the bark of the Pacific yew tree."^ Since its approval by the U.S. Food and Drug
Administration in 1992, taxol has been shown in clinical trials to be an effective treatment
for a number of cancers including breast, lung, and especially ovarian."' In fact in a
recent study, women suffering from advanced ovarian cancer who were given taxol in
combination with other anticancer medications lived an average of fourteen months longer
than patients who received other therapies."* Currently, researchers are pursuing the
challenge of creating whole families of synthetic taxol analogues which exhibit even better
therapeutic properties and can be used to treat a wider range of cancers."' Because of its
early success and fiiture potential, taxol is considered one of the most promising
treatments for cancer.
The mechanism of how taxol fiinctions in the human body was uncovered by
Horwitz and Schiff in 1978.'^" In their research, they found that taxol binds to tubulin, a
protein used to make structures in the cell known as microtubules. Microtubules serve as
part of the cell's internal skeleton and also play a crucial role in a number of vital functions
including cell division (mitosis). For a cell to divide, the microtubule skeleton must first
disassemble, then reform into spindle fibers which help to line up and separate the
duplicate sets of chromosomes. Once the DNA material is separated, the microtubules
132
must disassemble once more and reform into the skeletal systems for the two new cells.
The more traditional anticancer drugs such as vinca alkaloids and colchicine work by
tearing apart a spindle's microtubules so that the cancer cell cannot divide.'^' When taxol
attaches to tubulin, however, the protein loses its flexibility and the microtubules become
extremely stable and static. In this way, the microtubules can no longer disassemble and
the cancer cell is destroyed as it divides.'^" Because cancer cells divide more frequently
than healthy cells, taxol primarily attacks tumors which exhibit runaway cell division.
While the mechanism for taxol at the cellular level is well understood, there is still
a great deal of interest in how taxol actually reaches the cancer site and attacks the tumor
as a whole. One of the leading theories suggests that taxol concentrates in the vascular
network surrounding the tumor and attacks the outer shell of the tumor first. '^^ As the
tumor recedes, the vascular region also contracts allowing taxol to attack the next layer of
the tumor. This process is repeated until the tumor is destroyed. Other theories propose a
combination of processes in which taxol attacks the tumor from the outer and inner
regions simuhaneously.'^^ In an attempt to investigate these processes, MALDI was
performed on samples of ovarian tumor tissue from mice treated with taxol. The goals of
this experiment were twofold: first, to see if MALDI could be used to detect trace levels
of taxol from the complicated tumor tissue, and second to see if there was a preferential
location of taxol in the tumor.
MALDI MS and MS/MS of Standard Taxol
Pure taxol (M.W. 853) was obtained from Bristol-Myers Squibb. A standard 10"^
M solution was prepared by dissolving 0.01 g of the material in 25 mL of methanol then
133
making a 2.5:10 dilution. For MALDI, 1.0 ^xL of the standard solution was deposited
onto the sample probe and mixed with an equal volume of DHB matrix solution. After
drying, the sample was inserted into the ion source and the laser was fired at several places
on the sample until a satisfactory spectrum was obtained. Because the molecular weight
of taxol is greater than 650 Da, the mass range of the ion trap was extended by resonantly
ejecting the ions at a fi^equency of 130 kHz (qeject = 0.295). The MALDI MS spectrum
acquired for taxol after thirty laser shots is shown in Figure 4-9. Although the laser power
was adjusted to just above the threshold irradiance level, fragment ions were still observed
in the spectrum. This type of fragmentation is common with large, thermally labile
biomolecules, and generally results from metastable decay or collisions with the helium
buffer gas upon injection into the ion trap.^° To perform MS/MS on taxol, the (M+H)^
ion at m/z 854 was isolated using the forward-and-reverse scan method, then fragmented
by CID at q = 0.300 using an axial modulation frequency of 115 kHz. The scan fiinction
parameters used to obtain the MALDI MS/MS spectrum for taxol are listed in Table 4-5.
As can be seen in the MS/MS spectrum shown in Figure 4-10, abundant daughter ions
were produced from cleavage of the central ester linkage followed by successive losses of
acetic acid and benzoic acid groups.
134
100 200 300 400 500 600 700 800 900 1000
m/z
Figure 4-9. MALDI MS spectrum of taxol (M.W. 853) using DHB matrix.
135
Table 4-5. MALDI MS/MS scan function parameters used for the analysis of taxol.
(Axial Modulation)
Table
Volt.)
(DC
Volt.)
Freq.
Amp.
Time
#
Scan Table
Mass
Start q
End q
Start a
End a
(kHz)
(ms)
1
Reset
100
0.000
0.000
0.0042
0.0042
1
2
Preionize
100
0.400
0.400
0.0042
0.0042
0.5
3
Trig Laser 1
100
0.400
0.400
0.0042
0.0042
10
4
Cool
100
0.400
0.400
0.0042
0.0042
50
5
Trig Laser 2
100
0.400
0.400
0.0042
0.0042
10
6
Cool
100
0.400
0.400
0.0042
0.0042
50
7
Trig Laser 3
100
0.400
0.400
0.0042
0.0042
10
8
Cool
100
0.400
0.400
0.0042
0.0042
50
9
Trig Laser 4
100
0.400
0.400
0.0042
0.0042
10
10
0 400
0 400
0 0049
0 0047
—
—
so
11
Trig Laser 5
100
0.400
0.400
0.0042
0.0042
—
—
10
12
Cool
100
0.400
0.400
0.0042
0.0042
50
13
Trig Laser 6
100
0.400
0.400
0.0042
0.0042
10
14
Cool
100
0.400
0.400
0.0042
0.0042
50
15
Eject Low
332*
0.101
0.860
-.0014
-.0014
485
5
0.5
Masses
16
Eject High
^ ^
332
0.860
0.310
-.0014
-.0014
115
5
0.5
Masses
17
Cool
332
0.310
0.300
-.0014
-.0014
1
18
CID
332
0.300
0.300
-.0014
-.0014
115
10
10
19
Cool
332
0.300
0.091
-.0014
-.0014
1
20
Prescan
100
0.360
0.360
0.0042
0.0042
130
0
0.5
21
Scan
100
0.360
0.890
0.0042
0.0042
130
8
110
22
Empty Trap
100
0.000
0.000
0.0042
0.0042
1
* uncalibrated mass for taxol displayed on the monitor
136
500 -1
268
240
400 -
300 -
200 -
100 -
286
591
509
309
327
449
531
r
(M+H)
854
816
■551
798
-569
693
100
200 300 400 500 600 700 800 900
m/z
1000
Figure 4-10. MALDI MS/MS spectrum of a taxol standard after mass isolation and CID
of the (M+uy ion at m/z 854.
137
Preparation of the Ovarian Tumor Tissue
Three ovarian tumors approximately 10 mm in diameter were obtained from the
Bristol-Myers Squibb Oncology Division (Princeton, NJ). The tumors were human in
origin, but had been implanted subcutaneously as fragments into immunodeficient, nude
mice. Taxol was dissolved in a vehicle consisting of 10% cremphor, 10% ethanol, and
80% saline and administered to the mice intravenously. After approximately 1 hr. the
animals were sacrificed and the tumors were surgically excised using standard procedures.
Each of the tumors was then snap frozen, placed in a small plastic vial, and packed on dry
ice for shipment. Upon receipt at the University of Florida, the tumors were placed in
cold storage at -70° C. The concentration of taxol in the tumors was reported to be 10 -
50 ng/g of tumor.
In preparation for analysis one of the frozen tumors was cut in half, then sectioned
into thin slices approximately 0.5 mm thick using a disposable scalpel. Special care was
taken to ensure that both the inner and outer regions of the tumor were sampled in each
slice. Under microscopic observation, a reddish vascular region could be seen
surrounding the more whitish interior of the tumor. In contrast to the rat cerebral tissue
studied in the first application experiment, the surface of the tumor tissue was relatively
smooth.
MALDI Analysis of Ovarian Tumor Tissue
For the MALDI experiment, a thin section of tissue was cut down the center of the
tumor and placed onto the surface of the sample probe. The weight of the section was
138
previously measured to be 6.5 mg. After positioning the tissue on the probe tip, 5.0 |iL of
a 0.1 M DHB matrix solution in methanol was deposited dropwise onto the surface of the
tissue using an Eppendorf pipette. In our previous experiments with the rat cerebral
tissue, the matrix solution remained on the surface of the tissue for several minutes before
crystallizing. This was most likely due to the fact that the tissue had become saturated
during the incubation period. With the ovarian tumor, however, the majority of the matrix
solution was absorbed directly into the tissue. The sample was allowed to dry and
crystallize at room temperature. Observation of the sample under the microscope revealed
that matrix crystals had formed inside of the tissue as well as on the tissue surface. No
pockets or other inhomogenieties in the tissue surface were noted.
The top spectrum in Figure 4-11 shows the MALDI MS spectrum obtained for
taxol after isolating a 50 Da window around the (M+H)* ion at m/z 854 using the
forward-and-reverse scan method. Using the instrumental and scan fijnction variables for
the taxol standard, the isolated (M+H)"^ was resonantly excited and fragmented producing
the MS/MS spectrum shown at the bottom of Figure 4-11. Because of the relatively wide
isolation window used, the (M+Na)"^ peak at m/z 876 was present in both the MS and
MS/MS spectra. Each spectrum was acquired after thirty laser shots at an irradiance of
approximately lO' W/cm^. Comparison of the taxol daughter ion spectrum from the
ovarian tumor with the corresponding spectrum from the taxol standard showed good
agreement (Figure 4-12). After analysis, the tissue was removed from the ion source and
observed under the microscope. Inspection of the sample revealed that the laser had
burned completely through the tissue in several spots. The holes formed by the laser were
measured to be approximately 0. 1 mm in diameter. Based on the thickness of the section
139
Figure 4-11. MALDI MS (top) and MS/MS (bottom) spectra for taxol from a thin
section of ovarian tumor tissue obtained from a mouse treated with
taxol intravenously.
140
300 -I
250 -
200 -
c
0)
c
150 -
MS/MS of Taxol in
Ovarian Tumor Tissue
100 200 300 400 500 600 700 800 900 1000
m/z
MS/MS of a Taxol Standard
500 -,
400 -
300 -
— 200 -
100 -
(M+H)
854
100 200 300 400 500 600 700 800 900 1000
m/z
Figure 4-12. Comparison of the MALDI MS/MS spectrum for taxol in ovarian tumor
tissue (top) with the corresponding spectrum for a taxol standard (bottom).
141
(0.5 mm) the amount of tissue sampled was determined to be approximately 0.09%.
Assuming the maximum loading of the original tumor (50 ng/g), the spectra shown in
Figure 4-1 1 correspond to approximately 280 pg of taxol.
As was the case in the previous experiment with the cerebral tissue, the thick layer
of matrix crystals on top of the tissue prevented any correlation of the acquired spectra
with a specific region of the tissue. The lack of a microscopic viewing system in the
instrument compounded the problem. Experiments were repeated with laser desorption
alone (no matrix). However, no signal for taxol was observed. Finally, attempts were
made to analyze the different regions of the tumor separately by microdisecting the tumor
prior to adding the matrix solution. This too was unsuccessful due to the small size of the
tumor.
Analysis of Polymyxin from Human Plasma
In the final application experiment, MALDI was used to analyze the antibiotic drug
compound Polymyxin Bi. Polymyxins are cyclic amphipathic peptides containing free
amino acid groups and a fatty acid tail. Polymyxins are produced from Bacillus polymyxa
and are commonly used to treat a variety of bacterial diseases including pneumonia,
meningitis, and gonorrhea.*^ Polymyxins work by binding to the cell membranes of Gram-
negative bacteria and disrupting their structure and permeability properties. '^^ Recently
there has been a great deal of interest in developing liquid chromatography/tandem mass
spectrometric methods (LC/MS/MS) to quantitate polymyxin Bi in human plasma. While
these methods have been shown to produce adequate detection limits and accuracies, they
require lengthy extraction and separation procedures prior to analysis.* The goal of this
142
final experiment was to evaluate the use of MALDI as a quick screening method for
polymyxin Bi and potentially other drug compounds fi^om human plasma.
MALDI MS and MS/MS of Standard Polymyxin Bi
Polymyxin Bi (M.W. 1203) was obtained fi^om Sandoz Research Institute (East
Hanover, NJ). A standard solution was prepared by dissolving 1.0 mg of pure polymyxin
Bi in 10.0 mL of 50/50 methanol/water to give a final concentration of 8.3x10"' M. For
analysis, 2.0 \iL of the standard solution was first applied to the sample probe and allowed
to dry. Four microliters of a 1.0 M DHB matrix solution prepared in methanol was then
pipetted on top of the standard. The matrix solution dissolved the dried polymyxin Bi,
then crystallized upon evaporation of the solvent. The sample was fiirther dried under a
stream of warm air for an extended period up to 30 min. The MALDI MS/MS spectrum
for polymyxin Bi (Figure 4-13) was acquired using the instrumental and scan function
parameters used previously for taxol. The only modifications involved changing the table
mass of the isolation and MS/MS scan tables to correspond to the (M+H)* ion of
polymyxin Bi. MS/MS of the isolated (M+H)* ion of polymyxin Bi at m/z 1204 resuhed
in the production of several low intensity daughter ions. The major daughter ions at m/z
1 186, 1 168, and 1 150 resulted fi-om successive losses of water. The daughter ion at m/z
1103 resulted fi-om cleavage of the terminal acyl chain beta to the amide linkage.
Cleavage of the first and third amide bonds along the peptide tail resulted in the two
daughter ions at m/z 762 and m/z 963, respectively. A small peak at m/z 744 was also
produced fi"om loss of water fi-om the m/z 762 ion.
143
400 500 600 700 800 900 1000 1100 1200 1300
m/z
Figure 4-13. MALDI MS/MS spectrum of standard polymyxin Bi using DHB matrix.
144
MALDI Analysis of Human Plasma
Human blood was obtained from the University of Florida Infirmary and
centrifuged to obtain whole plasma. Immediately following, 1.0 mL of the plasma was
spiked with 1.0 mg of polymyxin Bi and vortexed for approximately 10 min. For MALDI,
1.0 |iL of the drug/plasma mixture was spread over the tip of the sample probe. One
microliter of DHB matrix solution prepared in methanol was then added on top of the
plasma mixture. The matrix solution dissolved some of the plasma sample resulting in a
gelatinous material on the probe tip. Irregular shaped matrix crystals were also observed
in the plasma as well as around the edges of the sample. The MALDI MS and MS/MS
spectra obtained for polymyxin Bi from the plasma sample after thirty laser shots is shown
in Figure 4-14. The entire analysis time from sample preparation to acquisition of the
spectrum took approximately 8 min. The LC/MS/MS method by Boue* was reported to
take close to an hour, including sample preparation. By comparing the MS/MS spectrum
from the plasma sample with the daughter ion spectrum obtained for the polymyxin Bi
standard (Figure 4-15), the presence of polymyxin Bi in the plasma was confirmed.
The shot-to-shot reproducibility and signal-to-noise (S/N) ratio for polymyxin
from the plasma were rather poor. Attempts were made to adjust the polarity of the
matrix solvent to extract more of the polymyxin out of the plasma. Since polymyxins
contain both hydrophilic and hydrophobic parts, the choice of solvent was challenging. In
the LC/MS/MS analysis of polymyxin Bi, solid phase extraction (SPE) cartridges were
used to extract polymyxin Bi from spiked plasma samples.* In that work, the plasma
145
300 -I
(M+H)*
1204
400 500 600 700 800 900
m/z
120 -I
100
80 -
(/)
c 60 -
40 -
20 -
MS/MS Of (M+H)
1000 1100 1200 1300
(M+H)*
1204
400 500 600 700 800 900 1000 1100 1200 1300
m/z
Figure 4-14. MALDI MS (top) and MS/MS (bottom) spectra of polymyxin B i from
a 1 .0 |iL sample of human plasma spiked with polymyxin Bj (1 mg/mL).
146
3000 -I
2500 -
2000
I 1500 H
1000 -
500 -
(M+H)*
1204
400 500 600 700 800 900 1000 1100 1200 1300
m/z
1186
1168
1150
662 963 1103-| I j
1
(M+H)*
1204
400 500 600 700 800 900 1000 1100 1200 1300
m/z
Figure 4-15. Comparison of the MS/MS spectra for polymyxin Bi mixed with human
plasma (top) and a polymyxin Bi standard (bottom).
147
samples were first loaded onto SPE cartridges. Polymyxin was then eluted fi-om the
cartridges using a series of solvents. The best recovery rates (60-67%) were obtained
using a 95% acetonitrile solution with 0.5% trifluoroacetic acid (TFA).
MALDI was repeated on a new 1.0 |jL sample of spiked plasma using a
concentrated DHB matrix solution prepared in 95% acetonitrile with 0.5% TFA. As was
the case in the previous MALDI experiment, it was difficult to obtain a reproducible ion
signal for polymyxin Bi fi-om the plasma, presumably due to the incomplete drying of the
plasma samples prior to analysis and the irregular matrix crystals that formed. In fiiture
experiments, a better approach to analyzing plasma samples by MALDI may be to
completely dissolve the analyte/plasma mixture first, then add in the matrix solution in a
similar solvent system prior to depositing the sample on the probe tip. Using this method,
the sample surface and the matrix crystal layer should be more uniform. While MALDI
may not be as quantitative an assay as LC/MS/MS, these initial results suggest that
MALDI does have potential as a fast screening method for drug compounds in plasma.
LD/CI as an Alternative to MALDI
From the results presented in this chapter, MALDI using a quadrupole ion trap
mass spectrometer was shown to be a viable technique for detecting trace levels of drug
compounds fi-om biological matrices. However, these same results revealed several
drawbacks to MALDI which may limit its potential for imaging drug compounds in tissue.
The first problem, discussed in chapter 2, is that the MALDI matrix solution can cause
migration of the analyte molecules from their original location in the tissue. Second, from
the experiments performed in this chapter with actual tissue from test animals, it was
148
learned that the matrix crystals can also make correlating the mass spectrum obtained from
the tissue with the actual spot sampled by the laser difficult at best.
An ideal alternative to MALDI for imaging compounds in tissue may be laser
desorption/chemical ionization (LD/CI). In LD/CI, the desorption laser is used to
vaporize the analyte into the gas phase as intact neutrals. Ionization of the analyte occurs
separately through ion molecule reactions (including charge exchange, proton-transfer,
and proton-abstraction) between the desorbed neutrals and an excess of ionized reagent
gas introduced into the ion source.'^'* Typical reagent gas pressures for CI are between
0.1-1.0 torr to ensure that there is a 1000-fold or greater excess of reagent gas in the ion
source. LD/CI was first introduced by Cotter'^^ in 1980 in experiments with a sector
instrument. In this work, reagent ions formed fi^om isobutane reagent gas were used to
ionize a series of glucuronide steroid neutrals formed by laser desorption. To date, most
of the applications of LD/CI have been for the analysis of low molecular weight peptides
deposited onto solid substrates. The analytical appeal of post-chemical ionization is that
the degree of fragmentation of the analyte can be controlled by the choice of reagent
ion.^" Also, since the yield of desorbed neutrals produced from laser desorption is
typically much greater than the yield of desorbed ions,'^* post-chemical ionization offers
the potential of enhancing the sensitivity significantly. The advantage of using LD/CI over
MALDI for imaging drug compounds in tissue is that there is no matrix solution added to
the sample which can cause the analytes to migrate.
The first application of LD/CI to the analysis of biological tissues was reported by
Perchalski at the University of Florida in 1985 using a Finnigan TSQ triple quadrupole
mass spectrometer.'^ The instrument was modified by machining the side flange of the
149
ion source to fit a fused silica lens used to introduce the desorption laser beam. The laser
used was a coaxial flashlamp-pumped dye laser operated with Rhodamine 6G. The
pulsewidth of the laser was 2.0 ^s with a pulse energy of 1.49 J. The EI/CI ion source of
the TSQ was pressurized with methane reagent gas (0.9 torr) and heated to 160° C.
Using this setup, LD/CI spectra were acquired for the antiepileptic drug phenytoin fi-om
fixed sections of rat liver tissue (Figure 4-16). The tissue was obtained fi^om a test animal
that had been given a loading of phenytoin (150 mg/kg) prior to sacrifice. In this work,
postchemical ionization was used to prolong the phenytoin ion signal lifetime up to 400 ms
in order to compensate for the slow scan speed of the quadrupole mass filter.
More recently in our laboratory, Vargas*^ attempted LD/CI of spiperone mixed
with matrigel using an external LDI source on a quadrupole ion trap instrument. The goal
of this experiment was to increase the sensitivity for detecting spiperone fi^om matrigel by
taking advantage of the excess of neutrals formed during the laser desorption process.
Samples were deposited onto the tip of a stainless steel probe and inserted into the
enclosed region of a high pressure ion volume situated in a Finnigan 4500 EI/CI ion
source. The laser beam fi-om a pulsed nitrogen laser was focused onto the sample surface
through a 0.125" diameter hole machined in the side of the ion volume. Using this setup,
however, no increase in ion signal was obtained for spiperone fi^om matrigel, presumably
due to the inability to achieve high enough reagent gas pressures in the ion volume to
perform efficient CI.
150
ION CURRENT (CNTS)
38848
175
182
210
253
225
^ IL
1087480 1
m/ z
175
182
180
210
225
253
200
220
240
260
Figure 4-16. Comparison of the LD/CI daughter ion spectra of m/z 253 for phenytoin
from rat liver tissue (top), and for phenytoin deposited onto a copper grid
(bottom).'^
151
Initial LD/CI Experiments with Trimethylphenylammonium Bromide
Before attempting LD/CI on analytes in tissue using the new laser desorption
instrument, experiments were first performed with a test compound,
trimethylphenylammonium bromide (TMPA). This compound was chosen to optimize the
laser desorption instrument for LD/CI because it absorbed strongly at the wavelength of
the desorption laser (337 nm) and therefore produced abundant ions and neutrals without
the need for MALDI. A concentrated solution (0. 1 M) was prepared by dissolving solid
TMPA in methanol. For analysis, 1.0 |iL of the standard solution was deposited onto the
surface of a stainless steel ion volume back. After drying, the CI ion volume cover was
placed over the sample and secured to the ion volume back with the small wire clip
(Figure 4-17). The entire ion volume assembly was then inserted into the ion source and
locked into place in fi^ont of the first extraction lens.
In the first experiment, laser desorption was performed on TMPA without the
presence of any reagent gas in the ion source. The laser beam was focused onto the
sample surface through the exit hole in the ion volume cover. LD spectra were acquired
using a modified version (lower qinject = 0. 1 80) of the original MALDI scan fimction for
spiperone. With laser desorption alone, the spectrum acquired for TMPA contained only
fragment ions (Figure 4-18). The peak at m/z 121 corresponded to loss of a methyl group
fi'om the TMPA ion; loss of a hydrogen from the m/z 121 ion gave the fi'agment peak at
m/z 120. A peak at m/z 59 was also seen in the spectrum resulting fi"om the loss of the
phenyl group fi'om the TMPA ion.
FILAMENT HOLE
Figure 4-17. Diagram of the CI ion volume assembly (left) used for the LD/CI
experiments. Samples were deposited onto the flat surface of the
ion volume back (right).i3°
153
154
In the second experiment, methane reagent gas was introduced directly into the
ion volume through a piece of 0.125" o.d. polyethylene tubing inserted through the GC
transfer port in the source block. Prior to triggering the laser, the rhenium filament in the
source was turned on to continuously ionize the reagent gas. The energy of the electron
beam was set to 70 eV by varying the potential difference between the filament and the
grounded source block. With the filament still on, the laser was fired at various spots on
the sample probe. Because ions formed by CI have significantly less initial kinetic energy
than MALDI generated ions, the voltage level on the source lenses had to be increased to
successfiilly extract and focus the ions fi^om the source into the DC quadrupole deflector.
The vohages on the DC quadrupole rods were changed only slightly. Initially, the spectra
acquired for TMPA were dominated by the fi-agment ions at m/z 121, 120, and 59.
However, by increasing the reagent gas pressure, the production of the molecular ion at
m/z 136 was improved significantly, as can be seen in Figure 4-19. The optimum pressure
in the ion source region of the vacuum chamber was measured at 6x10"^ torr from the
readout of the Bayard-Alpert ion gauge. The pressure inside the ion volume was
measured at 0.2 torr using a capacitance manometer. The increase in the molecular ion
signal for TMPA was due to a combination of collisional stabilization of the desorbed
analyte ions and charge exchange or proton transfer reactions between the desorbed
neutrals and the excess of methane reagent ions.
156
LD/CI of Spiperone in Rat Cerebral Tissue
Following the initial optimization experiments with TMPA, LD/CI was attempted
on a thin section of rat cerebral tissue that had been incubated in a solution of spiperone.
The sample preparation procedure followed was the same as that used previously for the
MALDI experiment, with the exception that no matrix solution was added to the tissue
prior to analysis. Also, instead of using the MALDI probe to introduce the sample into
the instrument, the tissue section was placed inside the enclosed ion volume. After
inserting the ion volume into the ion source, laser desorption was performed on the tissue
section. Without the addition of the MALDI matrix, no peaks were observed for spiperone
or for the tissue itself Following this experiment, methane reagent gas was introduced
into the ion source at an initial pressure of 0.2 torr. The laser was again fired at several
spots on the tissue. LD/CI was attempted for the tissue sample using reagent gas pressures
up to 0.9 torr. As was the case in the first experiment, however, no signal was recorded
for spiperone.
After analysis, the tissue sample was removed fi^om the ion source and observed
under the microscope. In contrast to the tissue samples analyzed by MALDI, there were
no laser holes in the tissue. The observations and results from these initial LD/CI
experiments suggest that without the addition of the UV matrix, the absorbance of the
tissue is not strong enough to allow the laser to probe through the tissue and desorb
analyte molecules embedded below the surface. This is not surprising given the relatively
low energy per pulse of the nitrogen laser (250 ^J/pulse). To overcome this problem in
the fiiture, a laser with significantly higher energy (1-2 J/pulse) will be needed.
CHAPTER 5
CONCLUSIONS AND FUTURE WORK
The goals of this project were to determine if MALDI could be used to detect
pharmaceutical drug compounds from intact biological tissues, and to evaluate the
potential of MALDI on a quadrupole ion trap mass spectrometer for future microprobe
analysis of drug compounds from complex biological matrices. In the first work with the
Lasermat instrument, MALDI was used successfully to increase the production of the
(M+H)"^ ion for spiperone and ephedrine from matrigel. The MALDI mechanism for drug
compounds in tissue was found to involve two sequential steps. In the first step, the
embedded analyte molecules were extracted from the tissue and into the matrix solution
applied on top of the tissue. In the second step, the extracted analyte molecules became
encapsulated in the matrix crystal layer which formed upon evaporation of the matrix
solvent.
Three parameters were found to be important in increasing the amount of analyte
detected using MALDI. Adjusting the polarity of the matrix solvent to match that of the
embedded drug compound was found to cause more of the analyte to partition out of the
tissue and into the matrix solution. In addition, the crystallization speed of the matrix
solution was found to have a significant effect on the production of analyte ion signal from
matrigel. It was observed that when the matrix droplet was allowed to soak on top of the
tissue for several minutes before crystallizing, a more efficient extraction of the drug
compounds from the matrigel was achieved. This in turn resulted in an increase in the
157
158
peak area of the analyte (M+H)* ion signal. The final parameter studied was the
concentration of the matrix solution. Addition of a small volume of DHB matrix solution
at concentrations of 0.01 M, 0.15 M, and 0.25 M resulted in the production of (M+H)"^
ion signals for both spiperone and ephedrine in matrigel. With the most concentrated
matrix solution, sevenfold and ninefold increases in the peak area of the (M+H)"^ ions of
ephedrine and spiperone were observed, respectively.
While the results of the experiments with matrigel clearly showed that optimizing
the MALDI process can increase the sensitivity for detecting drug compounds in tissue, it
is not suggested that lengthy optimization procedures be conducted for every potential
compound analyzed. Instead, future experiments should focus on developing MALDI
protocols for whole classes of compounds. For example, in this work three different drug
compounds (spiperone, taxol, and polymyxin Bi) were detected from biological matrices
using the same DHB matrix in methanol. The MALDI protocols should include the choice
of matrix compound, solvent mixture, and matrix solution concentration that work best
for a group of compounds. While DHB proved to be useful in all of the applications
presented here, it is likely that other MALDI matrices will perform better for different
classes of compounds. The use of co-matrices has been shown to increase the
reproducibility of MALDI signals and should therefore be investigated for analysis of
tissues." The advantage of developing MALDI protocols for whole classes of
compounds rather than individual compounds is that the analysis time would be
significantly reduced. These protocols would also be useful in cases were the tissue
sample being analyzed contains an unknown compound.
159
In the second part of this project, an external laser desorption ion source was
coupled to an ion trap mass spectrometer using a DC quadrupole deflector. The operation
of the ion injection system was optimized for MALDI using the ion optics simulation
program SIMION V6.0s. Experiments were performed to simulate the trajectories for ten
ions formed by EI and MALDI, respectively. The results of the simulations confirmed that
the DC quadrupole deflector functions as an energy analyzer. As the initial kinetic energy
of the was ions increased, higher voltage levels were required on the quadrupole rods of
the deflector to successfully turn and focus the ions into the ion trap. While simulations
were performed to investigate the trajectories for MALDI ions up to m/z 3000 (10 eV), a
full characterization of the DC quadrupole deflector system was not performed in this
work. Future simulation experiments should focus on determining the maximum m/z
MALDI ions that can be successfully turned and focused by the DC quadrupole deflector
using the current instrumental power supplies. Simulations should also be used to further
optimize the transmission efficiency of the ion injection system using greater numbers of
ions with the goal of increasing the overall sensitivity of the instrument
To evaluate the potential of the new instrument for future microprobe analysis,
three pharmaceutical drug compounds (spiperone, taxol, and polymyxin Bi) were analyzed
in tissues obtained directly from test species. In these experiments, the addition of the
MALDI matrix solution produced significant background ions from the tissue which
complicated the spectra. To overcome this problem, the (M+H)* ion for the respective
analytes was isolated inside of the ion trap, then fragmented by CID to produce daughter
ions. The presence of the drug compounds in the various tissues was confirmed by
comparing the resulting daughter ion spectra with the corresponding daughter ion spectra
160
for standards of each drug compound. The lowest level of drug compound detected was
280 pg for spiperone from rat cerebral tissue. The results of these experiments showed
that the MS/MS capabilities of the ion trap are necessary for detecting trace levels of drug
compounds from complex biological matrices.
While MALDI was shown to be a useful technique for detecting drug compounds
in tissue, it may prove less useful for applications involving imaging of analytes in tissue.
In the experiments with spiperone in matrigel, addition of the matrix solution caused the
analyte to migrate from its original location in the tissue. Electrospray deposition of the
matrix solution was performed and found to reduce the migration of spiperone in matrigel.
However, the sensitivity for detecting spiperone was significantly reduced because of the
rapid evaporation and crystallization of the matrix solution. Electrospray deposition was
not performed on any of the actual tissue samples from test animals prepared in this work.
Future experiments should be performed to determine whether electrospray deposition can
be used to detect trace levels of drug compounds from complex tissues. Another problem
encountered with MALDI was the fact that the thick matrix crystal layer that formed on
top of the tissue samples prevented specific regions of the tissue from being identified and
targeted for analysis. For tissue samples in which the location of the analyte of interest is
previously known, this problem can be overcome by cutting out the region of tissue
containing the analyte and analyzing it separately from the rest of the tissue. However, in
applications were the location of the drug is not known, it is necessary to be able to
observe the tissue surface in order to determine which regions of the tissue were sampled
by the desorption laser.
161
One possible solution is to apply the matrix solution to the sample probe first, then
place a thin tissue section on top of the dried matrix crystals. This technique is known as
substrate-assisted laser desorption ionization (SALDI). To perform SALDI of tissue
samples it is likely that a laser with at least a millijoule to possibly a joule of energy per
pulse will be needed to bum through the tissue to desorb both the embedded analyte
molecules and the matrix crystals underneath. Another possible alternative is to fi-eeze the
tissue samples and use the ice crystals that form within the tissue as the absorbing matrix.
The difficulty in doing this, however, is that the sample must be kept fi-ozen inside of the
vacuum chamber during the analysis. This can be achieved by introducing a liquid
nitrogen cold finger into the vacuum chamber in such a way that it makes contact with the
sample probe in the ion source. MALDI in this way would require an IR laser to match
the absorbance of the ice crystals.
One of the most promising alternatives for detecting drug compounds fi-om tissue
without the use of a matrix solution is laser desorption/chemical ionization. In this work,
preliminary LD/CI experiments were performed with TMPA using methane reagent gas.
By introducing an excess of reagent gas ions into the ion source during the laser
desorption event, the production of the intact molecular ion for TMPA was increased.
LD/CI experiments were also performed on samples of spiperone in rat cerebral tissue.
Without the addition of the UV-absorbing matrix, however, the laser beam was not able to
probe deeply enough into the tissue to desorb and ionize the embedded analyte molecules.
While the results of these initial experiments were inconclusive, they demonstrated that the
design of the new instrument was adequate for performing LD/CI. Future work with
LD/CI should focus first on understanding the LD/CI process itself To date, most of the
162
applications of post-chemical ionization have been made for thermally desorbed neutrals
from solids probes. Pressure studies and experiments with various reagent gases should
be conducted to optimize the CI process for laser desorbed neutrals. For LD/CI of tissues,
a more powerful laser (in the mJ-J/pulse range) should be employed to ensure that the
maximum number of analyte neutrals are desorbed from the tissue to increase the
sensitivity.
In order for the new laser desorption instrument to be used for microprobe
applications, instrumental improvements must be made to allow the sample to be viewed
under magnification inside of the vacuum chamber and to allow specific regions of tissue
samples to be targeted with high spatial resolution. With the ofF-axis design of the ion
source, samples can easily be viewed by positioning a microscope objective in front of the
ion source block. To secure the objective, a holder can be mounted directly to the DC
quadrupole deflector assembly. It is suggested that a reflecting microscope objective be
used since it incorporates special optics which can withstand the high irradiance levels of
the laser beam. These objectives are also ideal because they can provide magnification up
to 40x at focal lengths of greater than two inches. To view the image of the sample a
CCD camera and monitor can be positioned outside of the vacuum chamber. There are
two possible ways to pinpoint specific locations on samples which can be implemented on
the new laser desorption instrument. The first method involves rastering the laser beam
across the surface of the sample while holding the sample probe in position in the ion
source. This can be done by using a motor driven mirror positioned outside of the
vacuum chamber to focus the laser beam into the ion source. Motorized mirror systems
use a computer-controlled x,y,z micromanipulator to direct the laser beam with spatial
163
resolution as low as 0.1 |im. The advantage of this approach is that it requires no
modifications of the existing instrumental hardware. However, since the diameters of the
holes in the ion source lenses used to introduce the laser beam onto the sample are much
less than the diameter of the sample probe, the outer edges of tissues samples would not
interrogated by the laser beam. The second approach involves moving the sample probe
while keeping the laser beam fixed. This can be accomplished by adding an x,y,z
manipulator between the solids probe lock and the vacuum flange of the ion source. This
approach requires modifications to the source flange, but would allow the entire area of
the sample to be analyzed.
In addition to moving towards imaging of drug compounds in tissue, fiiture
applications of the laser desorption instrument should also focus on developing
quantitative assays for drug compounds in various biological matrices. One approach to
doing this is to use internal standards mixed with the matrix solution. Other potential
experiments include monitoring drug metabolites and possibly naturally occurring
biomolecules such as neurotransmitters directly fi^om biological tissues.
In conclusion, fi-om the results presented in this dissertation MALDI was shown to
be a an ideal technique for detecting drug compounds directly fi^om intact biological
tissues. It is also clear fi^om this work that the MS/MS capabilities of the ion trap are
needed for detecting trace levels of compounds fi^om very complex tissues. With the
addition of both a microscopic viewing system and a sample manipulation system, the new
laser desorption quadrupole ion trap instrument has the potential for a true molecular
microprobe.
LIST OF REFERENCES
1. Kahn, P. Science. 1995, 270, 368-370.
2. Umesh, V.; Welling, P. The Drug Development Process. Marcel Dekker Press: New
York, 1996, 447.
3. Nicholas, T.; Gardner, G; Gordon, D. Tetrahedron. 1995, 51, 8135-8173.
4. Thayer, A.M. Chem. Eng. News. 1996, 74, 7, 57-73.
5. Stahl, E. Drug Analysis by Chromatography and Microscopy. Ann Arbor Press:
Ann Arbor, 1973, 57-81.
6. Perun, T.J.; Propst, C.L. Nucleic Acid Targeted Drug Design. Marcel Dekker Press:
New York, 1992, 619-720.
7. Betzig, E.; Trautman, J.K. Science. 1992, 257, 189-195.
8. Boue. S. Ph.D. Dissertation. The Analysis of Antibiotics using Electrospray
Quadrupole Ion Trap Mass Spectrometry: IRMPD and LC/MS. University of
Florida, 1997.
9. Nowack, R. Science. 1995, 270, 369-371.
10. Williams, M.; Malick, J. Drug Discovery and Development. Humana Press: Clifton,
NJ, 1987,447-458.
11. Hercules, D.M.; Novak, F.P.; Viswanadham, S.K.; Wilk, Z.A. Anal. Chem. Acta.
1987, 193, 61-71.
12. Kelland, D.; Wallach, E.R. Anai Chem. Acta. 1987, 195, 89-96.
13. Lee, M.S.; Yost, R.A. Biomed. Environ. Mass Spectrom. 1988, 15, 193-205.
14. O'Shea, D.C.; Callen, W.R.; Rhodes, W.T. Introduction to Lasers and their
Applications. Addison Wesley: Menlo Park, 1978.
15. Honig, R.E.; Woolston, J R. Appl. Phys. Lett. 1963, 2, 138-139.
164
165
16. Conzemius, R.J.; Svec, H.J. Anal. Chem. 1978, 50, 1854-1861.
17. Bingham, R.A.; Salter, P.L. Anal. Chem. 1976, 48, 1735-1739.
18. Perchalski, R.J.; Wilder, B.J.; Yost, R.A. Anal. Chem. 1983, 55, 2002-2005.
19. Spengler, B.; Cotter, R.J. Anal. Chem. 1990, 62, 793-799.
20. Schwartz, J.C.; Bier, M.E. Rapid Commun. Mass Spectrom. 1993, 7, 27-32.
21. Verbuken, A.H.; Bruynseels, F.J.; Van Grieken, R.E. Biomed. Mass Spectrom.
1985, 12, 438-462.
22. Heinen, H.J.; Wechsung, R.; Vogt, H.; Hillenkamp, F.; Kaufmann, R. Biotechn.
Umschau 19S4, 2, 346-351.
23. Wechsung, R.; Hillenkamp, R.; Kaufmann, R.; Nitsche, R.; Vogt, H. Scanning
Electron Microsc. 1978, 1, 611-615.
24. Kaufmann, R.; Hillenkamp, P.; Wechsung, R.; Heinen, H.J.; Schurmann, M.
Scanning Electron Micros. 1979, II, 279-282.
25. Kaufinann, R.; Hillenkamp, F.; Wechsung, R. Med. Progr. Technol. 1979, 6, 109-
115.
26. Vandeputte, D.; Savory, J. Biomed. Environ. Mass Spectrom. 1989, 18, 598-602.
27. Laser Microprobe Mass Analyzer Brochure. Leybold-Heraeus: Export, PA, 1982.
28. Verbukueken, A.H.; Van Grieken, R.E. Anal. Chem. Acta. 1987, 195, 62-70.
29. Cotter, R.J.; Tabet, J. Amer. Lab. 1984, April, 86-99
30. Karas, M.; Bahr, U.; Giebmann, U. Mass Spectrom. Rev. 1991, 10, 335-357.
31. McCrery, D A; Gross, M.L. Anal. Chem. Acta. 1985, 178, 105-1 15.
32. Van Vaeck, L.; Claereboudt, J.; De Waele, J. Anal Chem. 1985, 57, 2944-295 1 .
33. Hillenkamp, F.; Karas, M.; Bachmann, D. Anal. Chem. 1985, 57, 2935-2939.
34. Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299-2301 .
166
35. Vertes, A.; Gijbels, R. Laser Ionization Mass Analysis. John Wiley & Sons: New
York, 1993, 127-149.
36. Karas, M.; Bahr, U.; Ingendoh, A.; Hillenkamp, F. Angew. Chem. Int. Ed. Engl.
1989, 28, 760-761.
37. Wu, K.J.; Steding, A.; Becker, C.H. Rapid Commun. Mass Spectrom. 1991, 5, 3 19-
324.
38. Stahl, B.; Steup, M.; Karas, M.; Hillenkamp, F. Anal Chem. 1991, 63, 1463-1472.
39. Hercules, D.M.; Kubis, A.J.; Somayajula, K.V.; Sharkey, A.G. Anal Chem. 1989,
61, 2516-2523.
40. Henzel, W.J.; Billed, T.M.; Stuhs, J.T.; Wong, S.C.; Grimley, C; Watanabe, C.
Proc. Natl. Acad Sci. U.S.A. 1993, 90, 501 1-5015.
41. Karas, M.; Bahr, U.; Ingendoh, A.I.; NordhofF, E.; Stahl, B.; Hillenkamp, F. Anal.
Chim. Acta. 1990, 241, 175-185.
42. Mock, K.K.; Sutton, C.W.; Cottrell, J.S. Rapid Commun. Mass Spectrom. 1992, 6,
233-238.
43. Eckerskom, C; Strupat, K.; Karas, M.; Hillenkamp, F.; Lottspeich, F.
Electrophoresis 1992, 13, 664-665.
44. Aebersold, R.M.; Teplow, D.B.; Hood, L.E.; Kent, S B. J. Biol Chem. 1986, 261,
4229-4238.
45. Fenselau, C; Vestling, M.M. Anal. Chem. 1994, 66, 471-477.
46. Strupat, K.; Karas, M.; Hillenkamp, F. Anal. Chem. 1994, 66, 464-470.
47. Beavis, R. Org Mass Spectrom. 1992, 27, 864-868.
48. Vertes, A.; Gijbels, R.; Irinyi, G. Anal Chem. 1993, 65, 2389-2393.
49. Spengler, B.; Kaufmann, R. Analusis. 1992, 20, 91-101 .
50. Wang, B.H.; Dreisewerd, K.; Bahr, U. Karas, M.; Hillenkamp, F. J. Am. Soc. Mass
Spectrom. 1993, 4, 393-398.
51. Borsen, K.O.; Schar, M.; Gassmann, E.; Steiner, V. Biol Mass Spectrom. 1991, 20,
471-478.
167
52. Beavis, R.C.; Chait, B.T.; Hillenkamp, F.; Karas, M. Anal. Chem. 1991, 24, 1193A-
1202 A.
53. Vorm, O.; Mann, M. J. Am. Soc. Mass Spectrom. 1994, 5, 955-958.
54. Caprioli, R.M. ; Andren, P.E. J. Mass Spectrom. 1995, 30, 8 1 7-824.
55. Hercules, D.M.; Proctor, A.; Wilkinson, W.R.; Gusev, A.I. Anal. Chem. 1995, 67,
1034-1041.
56. Williams, P.; Thomas, R.M.; Nelson, R.W. Rapid Commun. Mass Spectrom. 1990,
4, 348-351.
57. Beavis, R.C.; Chait, B.T. Rapid Commun. Mass Spectrom. 1989, 3, 432-435.
58. Zhao, S.; Somayajula, K.V.; Sharkey, A.G.; Hercules, D.M.; Hillenkamp, F.; Karas,
M; Ingendoh, A. Anal Chem. 1991, 63, 450-453.
59. Overberg, A.; Karas, M.; Bahr, U.; Kaufmann, R.; Hillenkamp, F. Rapid Commun.
Mass Spectrom. 1990, 4, 293-296.
60. Gimon, M.E.; Preston-Schafiler, L.M.; Russell, D. in Proceedings of the 42"'' ASMS
Conference on Mass Spectrometry and Allied Topics. Chicago, YL, 1994, 1347.
61. MALDI Matrices Reference Guide. Thermo Bioanalysis Limited: Hemel Hempstead,
UK, 1995.
62. Vargas, R. Ph.D. Dissertation. Development of Laser Desorption Ionization on a
Quadrupole Ion Trap Mass Spectrometer. University of Florida, 1993.
63. Raven, P.H.; Johnson, G.B. Biology. Times Mirror/Mosby PubHshing:St Louis,
1986, 906-908.
64. Lasermat Users Guide. Finnigan MAT: Hemel Hemstead, UK, 1991.
65. Winograd, N. Anal. Chem. 1993, 65, 622A-629A.
66. Booth, M M. Ph.D. Dissertation. Matrix-Assisted Laser Desorption Ionization
(MALDI) Tandem Mass Spectrometry (MS") Using a Quadrupole Ion Trap.
University of Florida, 1996.
67. Giddings, J.C. Unified Separation Science. John Wiley & Sons Inc.: New York,
1991, 29-30.
68. Cottrell, T.S.; Mock, K.K. Rapid Commun. Mass Spectrom. 1992, 6, 233-238.
168
69. Paul, W.; Steinwedel, H. U.S. Patent 2,939,952 filed December 21, 1954 (German
Patent filed December 1953).
70. March, R.E.; Todd, J.F. Practical Aspects of Ion Trap Mass Spectrometry Vol. I.
Fundamentals of Ion trap Mass Spectrometry. CRC Press: Boca Raton, 1995, 26-
48.
71. Paul, W.; Reinhard, H.P.; Zahn, U. Z Phys 1958, 152, 143-182.
72. Wuerker, R.F.; Shelton, H.; Langmuir, R.V. J. Appl. Phys 1959, 30, 342.
73. Major, F.G.; Dehmelt, H.G. Phys Rev. 1968, 179, 91-107.
74. Fischer, E. Z. Phys 1959, 156, 1-26.
75. Dawson, P.H.; Whetten, N.R. J. Vac. Sci. Techno!. 1968, 5, 1 1-18.
76. Stafford, G.C.; Kelley, P.E.; Bradford, D.C. Am. Lab. 1983, 51-57.
77. Weber-Grabau, M.; Kelley, P.E.; Syka, J.E.P.; Bradshaw, S C.; Hoekman, D.J. in
Proceedings of the 36** ASMS Conference on Mass Spectrometry and Allied Topics.
San Francisco, CA, 1988, 1106-1107.
78. Busch, K.L.; Glish, G.L.; McLuckey, S.A. Mass Spectrometry/Masss Spectrometry:
Techniques and Applications of Tandem Mass Spectrometry. VCH: New York,
1988, 87-90.
79. Kaiser, RE.; Louris, J.N.; Amy, J.W.; Cooks, R.G. Rapid Commun. Mass
Spectrom. 1989, 3, 115-119.
80. Yates, N.A.; Yost, R.A.; Bradshaw, S.C.; Tucker, D.B. in Proceedings of the 39^''
ASMS Conference on Mass Spectrometry and Allied Topics. Nashville, TN, 1991,
132-133.
81. Gronowska, I.W.; Paradisio, C; Traldi, P.; Vettori, U. Rapid Commun. Mass
Spectrom. 1990, 4, 306-313.
82. Kaiser, R.E.; Cooks, R.G.; Syka, J.E.P.; Stafford, G.C. Rapid Commun. Mass
Spectrom. 1990, 4, 30-35.
83. Duckworth, DC; Barschick, CM. Smith, D.H.; McLuckey, S.A. Anal. Chem.
1994,66,92-98.
169
84. Louris, J.N.; Brodbelt, J.S.; Kaiser, R.E.; Cooks, R.G. in Proceedings of the 35'''
ASMS Conference on Mass Spectrometry and Allied Topics. Denver, CO, 1987,
766-767.
85. Weber-Grabau, M; Kelley, P.E.; Syka, J.E.P.; Bradshaw, S.C.; Brodbelt, J.S. in
Proceedings of the 35'* ASMS Conference on Mass Spectrometry and Allied Topics.
Denver, CO, 1987, 769-770.
86. Van Berkel, F.J.; Glish, G.L.; McLuckey, S.A. Anal Chem. 1990, 62, 1284-1295.
87. Knight, R.D.; Int. J. Mass Spectrom. Ion Procceses. 1991, 106, 63-65.
88. Mathieu, E.J. Math Pure Appl 1868, 13,137.
89. Guidugli, F.; Traldi, P. Rapid Commun. Mass Spectrom. 1991, 5, 343-349.
90. Louris, J.; Stafford, G.; Syka, J.; Taylor, D. in Proceedings of the 40"' ASMS
Conference on Mass Spectrometry and Allied Topics. Washington, DC, 1992, 1003.
91. Johnson, J.V.; Pedder, R.E.; Yost, R.A. Rapid Commun. Mass Spectrom. 1992, 6,
760-764.
92. Kelly, P.E.; Hoekman, D.J.; Bradshaw, S C. in Proceedings of the 41" ASMS
Conference on Mass Spectrometry and Allied Topics. San Francisco, CA, 1993,
453.
93. Soni, M.H.; Cooks, R.G. Anal Chem. 1994, 66, 2488-2496.
94. Schwartz, J.C.; Jardine, I. Rapid Commun. Mass Spectrom. 1992, 6, 313-317.
95. Kaiser, R.E.; Cooks, R.G.; Stafford, G.C.; Syka, J.E.P.; Hemberger, P.H. Int. J.
Mass Spectrom. Ion Processes. 1991, 106, 79-85.
96. Cotter, R.; Heller, D.N.; Lys, I.; Uy, O.M. Anal. Chem. 1989, 61, 1083-1089.
97. Glish, G.L.; Goeringer, D.E.; McLuckey, S.A. Int. J. Mass Spectrom. Ion Processes.
1989,94, 15-24.
98. Vargas, R.; Yost, R.A. in Proceedings of the ASMS Sanibel Conference on Lasers
in Mass Spectrometry. Sanibel, FL, 1992.
99. Cotter, R.J.; Doroshenko, V.M.; Cornish, T.J. Rapid Commun. Mass Spectrom.
1992, 6, 753-757.
170
100. Vargas, R.; Yost, R.A. in Proceedings of the Pittsburgh Conference on Analytical
Chemistry and Applied Spectroscopy. Atlanta, GA, 1992, 548.
101. Beavis, R.C.; Chait, B.T. Chem. Phys. Letters. 1991, 181, 479.
102. Glish, G.L.; Chambers, D.M.; Goeringer, D.E.; McLuckey, S.A. in Proceedings of
the 4&'' ASMS Conference on Mass Spectrometry and Allied Topics. Washington,
DC, 1992, 1747-1748.
103. Louris, J.N.; Amy, J.W.; Ridley, T.Y.; Cooks, R.G. Int. J. Mass Spectrom. Ion
Processes. 1989, 88, 97.
104. Vargas, R.; Yost, R.A.; Lee, M.S.; Moon, S.; Rosenberg, I.E. in Proceedings of the
Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy. Atlanta,
GA, 1993, 1225.
105. Reddick, CD.; Booth, M M.; Yost, R.A. in Proceedings of the 44"" ASMS
Conference on Mass Spectrometry and Allied Topics. Portland, OR, 1988, 1366.
106. Zeman, H.D. Rev. Sci. Instrum. 1977, 48, 1079-1085.
107. Russell, D.H.; Hanson, D.H.; Castro, M E. in Proceedings of the SS"" ASMS
Conference on Mass Spectrometry and Allied Topics. Denver, CO, 1988, 1152-
1153.
108. Pedder, R.E.; Yost, R.A. in Proceedings of the 37"' ASMS Conference on Mass
Spectrometry and Allied Topics. Miami, FL, 1989, 468.
109. Pedder, R E. Ph.D. Dissertation. Fundamental Studies in a Qudrupole Ion Trap
Mass Spectrometer. University of Florida, 1992.
110. Griffin, T P. Ph.D. Dissertation. GC/MS/MS on the Quadrupole Ion Trap Mass
Spectrometer: Software Development and the Examination of the Effects of Ion
Population. University of Florida, 1995.
111. Yates, N.A. Ph.D. Dissertation. Methods for Gas Chromatography-Tandem Mass
Spectrometry on the Quadrupole Ion Trap. University of Florida, 1994.
112. Torrey, E.F. Surviving Schizophrenia. Harper Perenial: New York, 1995, 15-22.
113. Keefe, R.; Keefe, H. Understanding Schizophrenia: A Guide to New Research on
Causes and Treatment. The Free Press: Toronto, 1994, 53-72.
114. Borne, R.F. Drug Topics. 1994, October 10, 108.
171
115. Escandon, N.A.; Zimmermann, D.C.; McCall, R.B. J. Pharmacol. Exp. Ther. 1994,
268, 441-447.
1 16. Wani, M.C.; Taylor, H.L.; Wall, M.E. J. Am. Chem. Soc. 1971, 93, 2325-2327.
117. Forastiere, A.A.; Rowinsky, E.K.; Chaudry, V. Proc. of the Amer. Soc. of Clin.
One. 1 1. 1992, A-289, 117.
118. Rowinsky, E.K.; Donehower, R.C. Cancer Treat. Rev. 1993, 19, 63-78.
119. Nicolaou, K.C.; Guy, R.K.; Potier, P. Sci. Amer. 1996, June, 94-98.
120. Schiflf, P.B.; Horwitz, S.B. Biochemistry. 1981, 20(11), 3247-3252.
121. Lewis, R. Biophotonics. 1995, July/ August, 34-35.
122. Straubinger, R.M.; Sharma, A.; Mayhew, E. Second National Cancer Institute
Workshop on Taxol and Taxus. 1992, September 23-24, 115.
123. Fong, G.W.K.; Kho, B.T. J. Liq. Chrom. 1979, 2, 957-968.
124. Hart, K.J.; McLuckey, S.A.; Glish, G. J. Am. Soc. Mass Spectrom. 1992, 3, 549-
557.
125. Harrison, AG. Chemical Ionization Mass Spectrometry. CRC Press Inc.: Boca
Raton, 1983, 25-33.
126. Cotter, R.J. Anal. Chem. 1980, 52, \161-\11Q).
127. Speir, J.P.; Gorman, G.S.; Amster, I. J. Anal Chem. 1991, 63, 65-69.
128. van Breemen, R.B.; Snow, M.; Cotter, R.J. Int. J. Mass Spectrom. Ion Processes.
1983, 49, 35.
129. Perchalski, R. Ph.D. Dissertation. Characteristics and Application of a Laser
lonization/Evaporation Source for Tandem Mass Spectrometry. University of
Florida, 1985.
130. Finnigan Mat 4500 Series GC/MS Operator's Manual. Finnigan MAT: San Jose,
CA, 1981, 4-13.
BIOGRAPHICAL SKETCH
Christopher Damn Reddick was bom in Washington, DC, on April 19, 1970.
Shortly after his eighth birthday, his parents moved the family to the suburbs of Silver
Spring, Maryland. Growing up, Chris received most of his early education from staying
up late with his parents while they studied for their undergraduate and graduate degrees.
Because of this, Chris found his days in elementary school to be restrictive and rather
boring. To offset the drudgeries of school, Chris became active in music and sports. By
the age of thirteen he had learned to play several instruments in his junior high band and
was playing for two basketball teams, the county soccer team, and the local YMCA
football team. It should be pointed out that one of Chris's greatest achievements to date
was scoring four touchdowns to win the county football championship in the 95 lb. and
under division. The closest Chris came to science or engineering as a kid was dismantling
and rebuilding bicycles during his BMX racing days.
Although Chris was able to do well in school when it counted, he had yet to reach
his full academic potential. So in 1985, his parents enrolled him in St. John's Military
Academy in Washington, D C. Chris quickly adjusted and thrived in his new environment.
By his senior year, Chris had reached the rank of captain, was sixteenth in his class, and
was playing for his high school basketball team, which was ranked in the top ten in the
country. After graduation, he attended Carnegie Mellon University in Pittsburgh. Chris
initially started out as an aspiring music major, but quickly switched to the Math
172
173
Department. In his second semester he changed his major to chemistry because he
thought that this was a more promising career path, but more importantly because he
enjoyed tinkering in the lab. While at Carnegie Mellon, Chris interned at the Aluminum
Company of America (ALCOA) just outside of Pittsburgh. It was at ALCOA that he was
first introduced to analytical chemistry. Although his primary project involved developing
supercritical-fluid chromatography (SFC) methods for contaminants in process lubricants,
Chris realized early on that mass spectrometry was taking a commanding lead in analytical
chemistry. On the advice of his supervisor Dr. Jerry Marks (a fiiend of Rick Yost's) and
his mentor Dr. Robin Khosah, Chris applied to and was accepted at the University of
Florida in August of 1992.
While in graduate school, Chris focused his research on using laser desorption
ionization methods to analyze pharmaceutical drug compounds from biological tissues.
This research eventually lead him to construct a new laser desorption mass spectrometer
based on the quadrupole ion trap mass analyzer. For the final three years of his graduate
research, Chris was supported on a grant fi-om Bristol-Myers Squibb. Upon graduation,
Chris will continue work in the pharmaceutical area at Bristol-Myers Squibb in New
Brunswick, New Jersey.
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Richard A. Yost, Chair
Professor of Chemistry
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
mes D. Winefordner /
raduate Research Professor of
Chemistry
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is flilly adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
David H. Powell
Associate Scientist of Chemistry
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is flilly adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
James A. Deyrup ,
Professor of Chemistry
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
H^)ward M. Johnson ' /
Graduate Research Proressor of
Microbiology and Cell Science
This dissertation was submitted to the Graduate Faculty of the Department of
Chemistry in the College of Liberal Arts and Sciences and to the Graduate School and was
accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy.
August, 1997
Dean, Graduate School